Skip to main content

Full text of "Modern practice in the construction and maintenance of rail joints and bonds in electric railways. [2nd ed.]"

See other formats

A 11 105 lb3D7M 


A11 1021 63074 

/Technologic papers of the Bureau of Sta 
— *62:1920 C.1 NBS-PUB-C 1910 



l vfi&-dfenx*&K'*!<i 

| Hn«| 

■■:-:"--:'-: : .-.'>-::.r : .-';.-;:i-.; 

■;,'."ri iv'.\- •>-'"'.■" i.*^V,'*:-T-: ' 







#62, A «<* 

t %m^ 1 

Reference ty&k not to be 
taken from the Library. 


Technologic Papers 

of the 

Bureau of Standards 

S. W, STRATTON, Director 

No. 62 





E. R. SHEPARD, Associate Electrical Engineer 

Bureau of Standards 








Technologic Papers 

of THE 

Bureau of Standards 

S. W. STRATTON, Director 

No. 62 





E. R. SHEPARD, Associate Electrical Engineer 
Bureau of Standards 







Bureau of Standard* 
JUL 8 1926 



By E. R. Shepard 



I. Introduction 5 

II. Historical and general discussion of bonds and joints 7 

1. Bond requirements 8 

(a) Intimate and permanent contact with rail under service 

conditions 8 

(6) Durability 8 

(c) Ease of installation under service conditions 9 

(d) Low resistance 9 

(e) Proof against theft 9 

(/) Reasonable cost 9 

2. Types of bonds 10 

(a) Old types fc 10 

(6) Soldered bonds 11 

(c) Compressed terminal and pin terminal bonds 12 

(d) Brazed or welded bonds 15 

(e) Mechanically applied head bonds 17 

(/) Tubular bonds 18 

3. Cross bonding and special work bonding 18 

(a) Cross bonding. , 18 

(b) Special work bonding 19 

4. Welded and special joints 19 

(a) Cast weld 19 

(6) Thermite-welded joints 20 

(c) Electrically-welded joints 21 

(d) Arc-welded joints 21 

(e) The Nichols composite joint 22 

(/) Mechanical joints 22 

III. Compilation of information submitted by operating companies 22 

1. Questions submitted and nature of replies 22 

Rail bonds 23 

Welded and other types of rail joints 24 


4 Technologic Papers of the Bureau of Standards 

III. Compilation of information submitted by operating companies — Contd. p age . 

j. Compilation of data submitted by operating companies on bonds. . 27 

(a) Questions 1,2, and 3. Number and types of bonds 27 

(6) Question 4. Inspection of bonds 30 

(c) Question 5. Criterion for replacement 31 

(d) Question 6/ Average life of bonds 32 

(c) Question 7. Intervals for cross bonding 38 

(/) Question 8. Size of cross bonds 38 

(9) Questions 9 and 10. Bonding around special work 38 

(h) Question 1 1 . Theft of bonds 39 

(*) Question 12. Grade of labor for bonding. .... 39 

(;') Question 13. Bonding tools 40 

(k) Questions 14 and 15. Drilling 40 

(/) Question 16. Cost of bonding 40 

(m) Question 17. Causes of failure 45 

(n) Question 18. Resistance of bonds, new and old 47 

3. Compilation of data submitted by operating companies on welded 

and other types of rail joints 48 

(a) Question 1. Number of joints in use 49 

(b) Question 2 . Types of construction 50 

(c) Question 3. Electrical efficiency of joints 50 

(d) Questions 4 and 6. Life of joints and causes of failure 50 

(e) Question 5. Cost of joints 52 

(/) Question 7. Temperature variation in welded joints 54 

(g) Question 8. Expansion joints 55 

IV. Analysis of data 55 

1. The mechanical joint 55 

(a) Defective roadbed 56 

, (b) Nonuniformity of rail sections 56 

(c) Defective rail ends 56 

(d) Failure to grind joints 56 

(e) Loose bolts 57 

(/) Improved bolts 57 

(g) Joint plates : 59 

(h) Room for concealed bonds 59 

(*') Examples of bond failures 51 

(J) Special plates 62 

(k) Improved joint plates 62 

2. Types of bonds and features of installation 64 

(a) Comparison of compressed and pin terminal bonds 64 

(6) Stranded v. ribbon bonds 73 

(c) Use of solder and alloys with mechanically applied bonds. . 74 

(d) Mechanically applied head bonds 79 

(e) Electric-weld bonds 83 

(/) Oxy-acetylene welded bonds 88 

(g) Bonding of manganese and other special work 90 

(h) Bonding of converted steam roads 93 

(i) Double v. single bonding 94 

Rail Joints and Bonds 5 

IV. Analysis of data — Continued. 

2. Types of bonds and features of installation — Continued. 

(;') Economic and other considerations for the replacement of p a ge. 

bonds 102 

(k) Standards for replacement 104 

3. Welded and special joints 106 

(a) The cast weld 107 

(b) The thermite-welded joint 108 

(c) The electrically -welded joint no 

(d) The arc-welded joint 112 

(e) The Clark joint 117 

(J) The Nichols composite joint 117 

V. Experimental tests 1 18 

VI. General conclusions 121 

Bibliography 123 


While studying electrolysis and electrolysis mitigation during 
the past five years the subject of rail bonding and track conduc- 
tivity has been brought forcibly to our attention by observations 
on railways where electrolysis surveys have been made, by a vast 
amount of discussion of the subject in the technical press, and by 
conversation and correspondence with railway engineers. This 
keen interest in track bonding originated largely through the 
necessity of mitigating electrolysis, and while electrolysis continues 
to be the greatest stimulus for bond maintenance, it is as a rule 
justifiable solely from the standpoint of good operation, and is an 
absolute necessity for the successful operation of block signals. 

In view of the great variety of bonds and bonding practices in 
use at the present time, and the large percentage of failures after 
25 years of experimentation on the part of the operating and 
manufacturing companies, and after repeated calls for informa- 
tion and advice on the subject from railway engineers, the Bureau 
of Standards deemed it advisable to institute a thorough investi- 
gation regarding the present status of bonding and joint mainten- 
ance with the idea of disseminating information that will aid the 
companies in selecting bonds and joints, and in methods of apply- 
ing and installing them, as well as of calling attention to the im- 
portance of good track conductivity and its true relation to elec- 
trolysis and its effect upon electric railway operation. 

6 Technologic Papers of the Bureau of Standards 

Owing to the peculiar nature of the services that rail bonds and 
joints are called upon to perform and the great variety of condi- 
tions under which they operate, it was recognized that informa- 
tion obtained under service conditions over a period of years 
would be far more reliable and satisfactory than any laboratory 
or short-time tests that could be conducted. Many laboratory 
tests have been made by the manufacturers on the durability and 
resistance of different types of bonds and joints, and while such 
tests are valuable in determining the characteristics of a bond, 
they can not be taken as a criterion for the performance of the 
average bond under service conditions. 

The personal element which enters so largely into the installa- 
tion of bonds, and the variety of conditions under which they op- 
erate made it necessary to obtain information from a large num- 
ber of sources and to base conclusions only on testimony sub- 
mitted by a great many witnesses. Accordingly, data were col- 
lected through a large number of circular letters and other corre- 
spondence, by personal visits to some 50 operating companies, as 
well as to practically all of the manufacturers of bonds and rail 

Owing to the rapid growth of the electric railways, the numerous 
changes in the standards of construction, the improvements in 
materials and methods, the franchise and street-paving require- 
ments, the changes in organization and administration, and the 
transient nature of the engineering staffs, it was found difficult to 
get definite and consistent information regarding the operation of 
any type of bond or joint over a period of years. Many of the 
companies consulted had kept only meager, if any, records of 
their bonding, and their replies, of necessity, were based largely 
on opinions. Few engineers were able to give with any degree of 
certainty the average life of a given type of bond. Either the 
bond had not been in service long enough to warrant a statement 
or else no definite records were available. 

New inventions and recent improvements in the manufacture 
and installation of bonds and joints, as well as changes in the 
types and composition of rails, all contribute to an unsettled con- 
dition at the present time, which means that a great number of 
bonds and joints now in service are in the experimental stage. 

Rail Joints and Bonds 7 

While the larger operating companies have the engineers and 
resources with which to meet the bonding problem with more or 
less success, the smaller companies must depend in a large measure 
upon these and other external sources for their standards. This 
has not always been to their advantage, as many practices em- 
ployed by the larger companies are not applicable to the smaller 
systems. Local conditions or city restrictions frequently limit 
the types of bonds that might be employed. The result is that 
many companies have been confused by the apparent inconsis- 
tencies in adopted standards, not realizing that each has been 
worked out to meet peculiar conditions. 

A great many railways have settled on certain bonding stand- 
ards and are apparently satisfied with the results they are obtain- 
ing, not that they believe they have the one and only best stand- 
ard, but they are tired of experimenting and are willing to let well 
enough alone. Others, and perhaps the majority of the com- 
panies, are not satisfied with their present practices, and are 
looking for something better suited to their requirements. 

In view of the conditions described above it can not be expected 
that this investigation will clear up a most vexatious question, 
nor in any way purport to be the last word on the subject. The 
Bureau of Standards hopes, however, in tabulating and analyzing 
the data which have been collected, to discern and interpret the 
present tendencies and to reconcile some apparent inconsistencies 
and differences of opinion. Its aim will have been fulfilled if it 
succeeds in laying before the electric railway companies, and par- 
ticularly the smaller companies with limited resources, informa- 
tion which will be a guide to the selection of bonds and joints; 
and, what is of still more importance, in pointing out the best 
methods of application and maintenance and in emphasizing the 
necessity of adhering to them. 



Although early attempts were made to operate cars on unbonded 
tracks, relying upon the joint plates and the earth for conductance, 
it soon became evident that a metallic bond was necessary both 
from the standpoint of good operation and for the prevention of 

8 Technologic Papers of the Bureau of Standards 

stray currents, which were early found to have a corrosive action 
on underground structures. 

Numerous types of iron and copper bonds came into use, many 
of which are now obsolete, and a description of which would be 
of no particular value to this paper. Those types which have 
found general use will be discussed with reference to their features 
of installation and the conditions under which they operate. No 
academic classification of bonds will be attempted. 

The great number of different types of bonds which have 
appeared in the past years is largely the result of attempts to 
better meet the exacting requirements which this piece of appa- 
ratus is called upon to fulfill. 

While these requirements and the different types of bonds 
which they have called forth are familiar to the majority of rail- 
way engineers, the most important of them are here briefly 
described by way of introduction to a later part of the paper, 
where, in connection with testimony submitted by the operat- 
ing and maufacturing companies, a more detailed account of the 
manner in which the various types of bonds are meeting the 
requirements of service will be discussed. 


(a) Intimate and Permanent Contact With Rail Under 
Service Conditions. — Perhaps the first and most important 
requirement of a rail bond is that it make good electrical contact 
with the rail and that this contact remain good over a period of 
years while subjected to the mechanical vibrations of traffic, 
changes of temperature, the action of soil and moisture, and to 
the mechanical injuries from workmen and vehicles. In general, 
three methods of making contact with the rail have been employed, 
viz, soldering, mechanical, and brazing or wielding. Combina- 
tions of these have also been used. Each method will be treated 
under the types of bonds employing it. 

(6) Durability. — The durability of the bond itself depends, 
first, upon its ability to withstand the bending and vibration 
incident to expansion and contraction of rails and the deflection 
of the rail joint under traffic, and, second, upon its ability to 
withstand electrolytic and soil corrosion. 

Rail Joints and Bonds 9 

The first action is by far the more severe and has resulted in the 
failure of more bonds than has any other one cause. Loose rail 
joints are the chief cause of such failures, and the problem of 
bonding is therefore intimately related to the problem of joint 

The second cause of deterioration, that of corrosion, is rarely 
important, although in extreme cases it may be serious. Iron 
bonds buried in the earth and copper bonds in soils of certain 
character, or on tracks from which large leakage currents are 
escaping, have been known to corrode at a very rapid rate. 

(c) Ease of Installation Under Service Conditions. — For 
the greatest practical value a bond should be of such a nature 
that it can be safely and quickly installed under service conditions ; 
that is, while traffic is being maintained over the tracks. For 
new work this may not be required, but for repair and replacement 
work its importance is obvious. While bonds are frequently 
installed at night and traffic is sometimes diverted for the purpose 
of installing bonds, such practices are decidedly objectionable. 
Moreover, it is possible for bonding apparatus to offer a hazard 
to the safe operation of cars. Though not of prime importance, 
these features can not be neglected in the selection of a bond. 

(d) Low Resistance. — Under ordinary conditions the resist- 
ance of a bond within moderate limits is subordinate to its other 
qualities. Obviously, it must have a cross section sufficiently 
large to carry the track current without undue heating, but as a 
rule its length will be determined by other considerations. Where 
the resistance of bonded joints is limited by ordinance, or where 
for special reasons a high conductance is required, this feature 
may be a determining factor in the selection of a bond. 

(e) Proof Against Theft. — In many localities the theft of rail 
bonds has become so prevalent and the losses from this source so 
heavy that the resources of the railways have been seriously taxed 
to cope with the problem. To-day no bonding of suburban track 
can be undertaken without due consideration of this feature, and 
either bonds designed to overcome this trouble must be selected or 
some other preventive means employed. 

(/) Reasonable Cost. — While the consideration of cost can not 
be neglected in the selection of apparatus and material, its promi- 

io Technologic Papers of the Bureau of Standards 

nence may in some instances be entirely overestimated. It is not 
always apparent that ultimate economy may result from a high 
first cost, and with the difficulty in securing approval of estimates 
for bonding from those who do not always appreciate the impor- 
tance and necessity of this work the more expensive bond is likely 
to be seriously handicapped in its bid for consideration. Although 
the general manager who wants to know why bonding is necessary 
when the cars are operating under existing conditions is rather the 
exception, there are many who place bonding at the end of the 
budget, so that the engineer is sometimes forced to employ material 
and methods for the sake of economy which are against his better 
judgment. It is quite evident, therefore, that the first cost of a 
bond, which of course includes the cost of installation, although 
having little relation to its ultimate economy, in many cases might 
be the determining feature in its selection. 


(a) Old Types. — Early bonding was accomplished by riveting 
or bolting solid iron or copper wires to the web or base of the rails. 
This general practice was not long employed, as it was soon found 
that such contacts rapidly deteriorated from corrosion. Practi- 
cally none of these bonds are in use at the present time and the 
types can be considered as obsolete. 

Channel-pin bonding, as shown in Fig. i, consists in driving 
a grooved plug into a hole in the rail with a round wire fitted into 
the groove in the plug, found early favor with the railway com- 
panies owing to its low cost and ease of installation. It is still 
to be found in service on old tracks and has a very limited sale at 
the present time for temporary use in mines and for other special 
purposes. These special conditions, however, are being met by 
modern and more satisfactory types, thus leaving this bond with 
a very limited field. 

Originally steel plugs and solid copper wires were used, but the 
practice has been varied by the use of copper instead of steel plugs 
and in other cases by copper-plated or tinned plugs. Sometimes 
the plug entirely encircles the wire in the form of a sleeve. 

Although there are cases on record where this type of bond 
maintained good electrical contact with the rail for many years 

Bureau of Standards Technologic Paper No. 62 

Fig. i. — Riveted and channel pin bonds 

Fig. 2. — Compressed terminal type 

Rail Joints and Bonds 1 1 

under service conditions, the results obtained in general were poor. 
Moisture invariably found its way between the plug and the rail, 
or between the wire and the plug, thereby causing corrosion and 
an increasing contact resistance. 

Channel pins and iron bonds still find a limited application on 
suburban tracks where the theft of copper bonds excludes that 
type, and where the rail and joint plates are of such dimensions as 
not to permit the use of concealed bonds. Under such conditions 
they are admittedly a makeshift and are employed only as a last 
resort and in the absence of any satisfactory method of bonding. 

(b) Soldered Bonds. — With the failure of the riveted, bolted, 
and channel-pin bonds the necessity of a bond making a more 
perfect and permanent contact with the rail became apparent. 
The soldered contact early came into use to meet this demand and 
found universal adoption. With the exception of the most modern 
installations practically every electric railway company in the 
country has employed the soldered bond in one form or another. 
Its low cost and ease of application were in its favor and appealed 
to the operating companies. It can be applied to either the head, 
web, or base of the rail, requires no drilling, and can be installed 
without interruption to traffic. 

The one serious objection to this type of bond is the difficulty in 
securing a permanent and low-resistance contact. The failures of 
soldered contacts are due to inherent defects in the method as well 
as to poor workmanship in installation. Copper has a coefficient 
of expansion nearly twice that of steel and somewhat less than 
lead-tin solder. It is evident, therefore, that with the diurnal 
temperature variations that steel rails undergo the soft film of 
solder connecting the two different metals is subjected to continual 
alternate strains which, in the presence of moisture and under the 
vibrations due to traffic, will eventually result in failure. 

There are few mechanical processes in which the personal ele- 
ment enters so largely as in the application of bonds, and this is 
particularly true with respect to the soldered bond. As a rule, 
skilled mechanics are not emp^ed for this work and the ordi- 
nary track laborer is slow in mastering the apparently simple feat 
of soldering a rail bond. In fact, he is a rare workman if he ever 
does learn the intricacies of this process and conscientiously ap- 

12 Technologic Papers of the Bureau of Standards 

plies his knowledge at all times. The soldered contact between 
bond and rail frequently corrodes without exhibiting any external 
signs of such deterioration. The bond might even resist a mod- 
erate blow from a hammer and still show a very high resistance 
when tested with a bond tester. Inspection therefore is not a 
reliable means of determining the condition of soldered bonds. 

Another inherent defect of the soldered bond is the compara- 
tive ease with which it can be removed from the rail by copper 
thieves. A short bar is all that is necessary to remove these 
bonds from the head of a rail, and enormous losses of this type 
have occurred where the labor and time necessary to remove other 
types would have saved them. 

While there are some engineers who still retain faith in the 
soldered bond, and a few companies employing well-trained and 
careful workmen continue to install them, their inability in gen- 
eral to meet the requirements of service have led to their abandon- 
ment by the majority of operating companies. 

(c) Compressed Terminal and Pin-Terminal Bonds. — Com- 
pressed terminal bonds, shown in Figs. 2 and 3, are those having 
cylindrical terminals which are compressed with a screw or hy- 
draulic compressor into holes drilled or punched in the web or 
base of the rails. They are referred to by the various manufac- 
turers as compressed terminal, solid terminal, and compressed 
stud terminal bonds. Pin-terminal bonds are those having tubu- 
lar terminals which are expanded into holes drilled or punched in 
the web or base of the rail by driving a steel pin into the hole in 
the terminal. They are referred to by the various manufacturers 
as pin-terminal, tubular-terminal, and pin-driven bonds. They 
will be referred to in this paper simply as pin-terminal bonds. 
The term "stud terminal" will be used to include both of the 
above types. 

These bonds are made either in the solid, stranded, or ribbon 
type, and are designed either for concealed or exposed applica- 
tion. Although a great diversity of opinion exists regarding the 
merits of these bonds, they have found wide application and for 
more than 15 years have remained the standards for numerous 
companies. Owing to their wide use at the present time and in 
view of the general interest manifested by the companies in them, 

Bureau of Standards Technologic Paper No. 62 

Fig. 3. — Pin-terminal bonds 

Rail Joints and Bonds 13 

a somewhat detailed account of their properties will here be in 

Some difficulty was at first experienced by the manufacturers 
in securing a perfect union between the terminal and the strands 
or ribbons of these bonds. The successful welding of copper re- 
quires certain precautions in the exclusion of oxygen, and until 
improvements were made in their furnaces and methods the bond 
manufacturers found difficulty in turning out bonds with properly 
welded terminals. This defect has been entirely overcome, so 
that it is now possible to obtain bonds which are perfect in this 
respect. To overcome this trouble some manufacturers have 
forged the terminals from the wire strand of the bond itself. 

Another improvement in the construction of stud-terminal 
bonds which has come into use only in recent years is that of ma- 
chining the terminals. Before this practice was employed the in- 
equalities in the stud made it difficult to obtain a perfect contact 
over the entire surface of the terminal. This gave a chance for 
the entrance of moisture between the copper and steel, with re- 
sulting corrosion and rapid deterioration. A film of moisture on 
the contact acts as an electrolyte, and with the passage of electric 
current rapid corrosion took place. A number of manufacturers 
now machine all bond terminals, while others do so only when 
specifications require it. The additional expense is small and 
most companies are willing to meet it for the increased life of the 
bond. Other improvements in the nature of annealing or soften- 
ing the copper have added to the value of stud-terminal bonds by 
permitting a better flow of copper and consequently a better union 
with the steel. 

Failure of the bond itself, due to the crystallizing and breaking 
of the wires and ribbons, has been largely overcome by increasing 
its length and by using a size of wires which experience has shown 
will withstand the maximum amount of mechanical vibration. 
Attention has also been given to the matter of forming bonds to 
conform to the joint plate and rail sections. 

While the manufacturers have been active in their efforts to 
reduce the failures of the stud-terminal bonds, by introducing 
improvements and refinements in their methods of construction, 
the utility of these, as well as all other types of bonds, has been 

1 4 Technologic Papers of the Bureau of Standards 

greatly increased by improved methods in their installation. The 
importance of great care in the installation of stud-terminal bonds 
was, at first, not always appreciated by either the engineer or the 
workman. The many precautions and refinements now known to 
be imperative for best results were not known in the early days of 
electric railway engineering. Electric roads were springing up 
like mushrooms in every city. In many cases they were built 
by contract at so much per mile and concealed bonds received 
scanty attention. Indeed, cases are on record where bonds have 
been removed from such roads after a period of years and were 
found not to have been compressed or expanded at all but simply 
driven in the holes drilled to receive them and covered up by the 
joint plates. Bonds poorly installed frequently gave good service 
for a short time ; sometimes for a few years. Even if they did not, 
the joint plates and the earth roadbeds frequently sufficed to re- 
turn the current to the generators, and until water and gas 4eaks 
began to develop, or until the necessity for better service called 
attention to the poor return circuit, the bonds often received no 
consideration. It is little wonder, therefore, that years were re- 
quired to establish the facts regarding the proper methods of 
bonding. The high percentages of failures that have been recorded 
in the past were apparently, therefore, on bonds which were manu- 
factured and installed under conditions far different from those 
existing at the present time or even in recent years, and they can 
not be considered as an index to the performance of modern bonds 
installed under more favorable conditions. 

Some of the features of installation which have contributed to 
the failure of stud- terminal bonds are here recounted: (i) The 
drilling of too large holes, thus requiring too great compression 
or expansion to make good contact. Holes are now usually 
drilled having the same diameter as that of the bond terminals. 
(2) Rough or irregular holes caused by dull or imperfectly ground 
drills. (3) Installing bonds in old, corroded, or wet holes. Bonds 
are now installed in only freshly drilled holes, perfectly clean and 
dry. (4) The use of oil in drilling. A film of oil between bond 
terminal and steel impairs the contact. Holes are now, as a rule, 
drilled dry. (5) Failure to clean web of rail around hole. Rust 
or scale on the web of the rail with which the face and button of 

Rail Joints and Bonds 15 

the bond comes in contact is likely to permit the admission of 
moisture to the contact. The best modern practice requires the 
grinding of the rail adjacent to the hole. (6) Incomplete com- 
pression or expansion of terminals. (7) Old or wrongly shaped 
compressor face. (8) Carelessness in driving expanding mandrel 
or pin. (9) Failure to clean bond terminal before installation. 

As with the soldered type the personal element enters largely 
into the installation of stud-terminal bonds, evidence of which is 
given by reference to the numerous details recounted above. 

The foregoing is considered a sufficient introduction to the 
discussion and comparison of the pin terminal and compressed- 
terminal bond which is to follow in connection with the reports of 
the operating companies. 

(d) Brazed or Weeded Bonds. — This type includes all bonds 
in which either the copper terminal of the bond is welded directly 
to the rail or in which a third metal, such as brass or some other 
hard solder, is used to effect the union. Heat may be applied by 
any means, the most common being the passage of an electric 
current through the members being united, the electric arc, and 
the oxy-acetylene flame. The pouring of molten copper into a 
mold around the bond terminal has also been employed. The 
following definitions are in common use and will be adhered to in 
this paper. 

Electric Weld. — Though commonly called a brazing process this 
term is used by the Electric Railway Improvement Co. in reference 
to the operation in which current passing through carbon electrodes 
in contact with the bond terminal generates the welding heat. It 
will here be employed in that connection. 

Arc Weld. — A weld or brazing process affected by heat gener- 
ated by an electric arc. 

Oxy- Acetylene Weld. — A weld or brazing process affected by 
the use of the oxy-acetylene flame. 

Copper Weld. — A weld or brazing process affected by pouring 
molten copper into a mold surrounding the bond terminal. 

Up to the present time the use of the welded or brazed bonds has 
been confined almost entirely to that of the electric- weld type. 

These bonds are more modern than the soldered and mechani- 
cally applied types described above, and their manufacture and 

1 6 Technologic Papers of the Bureau of Standards 

use has been greatly stimulated by the high percentage of failures 
attributed to the imperfect contacts of the latter-named types. 
A greater stimulus, however, was the growing need for a short 
exposed bond which could be applied to the head of the rail 
without removing the joint plates and which would make such 
tenacious contact as to discourage the attempts of copper thieves 
to remove it. The need for such a bond was so strong and the 
brazed or welded bond met the requirements so admirably that it 
at once sprang into extensive use, particularly on open track, 
even before time had demonstrated its lasting qualities. To 
guard against theft on suburban tracks and also to reduce its 
cost the bond was of necessity made comparatively short. This 
feature led to a rather high rate of failure from the breakage of the 
wires or ribbons, particularly on roads having poorly maintained 
rail joints. Imperfect methods of application and carelessness 
in installation also contributed to the failures of this type. 

Like the stud- terminal bonds, therefore, the electric-weld bond 
had to go through an experimental stage. Improvements in 
construction, adoption of new types, as well as education of the 
railways in the methods of installation have progressed until 
to-day most of the early defects have been overcome and the bond 
is supplying a wide and growing demand. 

The chief objections to the electric-weld bond are, first, as 
usually applied, their installation requires the purchase of rather 
an expensive bonding car, and, second, the bonding car is incon- 
venient to operate on tracks over which traffic is being main- 
tained. The first objection is a serious one for small roads having 
limited means, and the second objection applies to all tracks on 
which the headway is of the order of 30 minutes or less. It is 
necessary to derail the bonding car to let regular trains pass, and 
this is obviously impracticable under a short headway. These 
objections are not pertinent to other types of welded bonds, and 
for this reason they are now making a strong bid for recognition. 

Figs. 4 and 5 show two types of electric-weld bonds in common 
use. The ET type, which is the newer of the two, was recently 
designed to overcome defects in the EA type The manner in 
which the conductors of the EA type met and joined the bond 
terminal permitted of considerable bending and vibration at that 

Bureau of Standards Technologic Paper No. 62 

Fig. 6. — Car for applying electric-weld bonds 

Fig. 6a.— Portable welding transformer 

Bure.-.. No. 62 

Fig. 7. — Twin terminal bond 


Fig. 8. — Tubular terminal bond 


Fig. 9. — Thermite weld 

Rail Joints and Bonds 


point on poorly maintained joints and resulted in considerable 
crystallization and breaking of the ribbons. In the ET design 
the flexure is distributed throughout the entire length of the bond, 
thereby greatly reducing the failures from this source. 

Where alternating current is available a portable outfit, shown 
in Fig. 6a, is employed. It is less expensive and more convenient 
to use than the bonding car. 

(e) Mechanically Applied Head Bonds. — Under this heading 
will be included the twin-terminal bond shown in Fig. 7 and the 

Fig. 4. — Type EA electric-weld bond 


Fig. 5. — Type ET electric-weld bond 

tubular-terminal bond shown in Fig. 8, both of which are short 
bonds applied to the head of the rail and used principally on 
suburban open track. The former employs two studs on each 
terminal about 1 % inches apart which are driven into holes drilled 
in the head of the rail and upset, while the terminal of the latter 
type is tubular in form and is driven into an annular shaped 
hole milled in the head of the rail. Both types are short, compara- 
tively inexpensive to install, make fairly good contact with the rail, 
can easily be inspected, and are relatively hard to steal. They 
may still be said to be in the experimental stage, however, and a 

150211°— 19 2 

1 8 Technologic Papers of the Bureau of Standards 

few more years will be required to demonstrate their ultimate 

(/) Tubular Bonds. — Recently a tubular copper bond has 
been placed on the market which is designed to join the web of 
the rail with the joint plates by expanding it into holes drilled 
through the three members at right angles to the length of the rail. 
It is about 4 inches in length and \% inches in external diameter. 
At least two are required for each joint and more may be used 
This bond has not been in service long enough to demonstrate its 
utility, but as it is called upon to bear the joint strains incident 
to expansion and contraction its ability to maintain good contact 
with the rail and plates is very questionable. However, if installed 
on improved mechanical joints where bolt holes are reamed for a 
driving fit, it is possible that this bond would prove entirely satis- 


(a) Cross Bonding. — If all joint bonding could be maintained 
in perfect condition there would be no occasion for cross bonding 
of the two rails of a single track, although on double track the 
utility of cross bonding between tracks is obvious. As perfect 
bonding is an ideal condition, practically never realized, experience 
has shown the necessity of cross bonding both single and double 
track at frequent intervals. The function of cross bonds on single 
track is to shunt the current around poorly bonded or high- 
resistance joints, thereby tending to equalize the return current in 
the two rails. On double tracks they not only act as shunts 
around high-resistance joints but also act as equalizers between 
tracks which may not always be uniformly loaded. In the absence 
of cross bonds serious unbalancing of the return current is likely to 
result on poorly bonded tracks, which not only makes for excessive 
power loss and poor operation but introduces bad electrolysis con- 
ditions. Under extreme conditions it may also actually prove a 
life hazard to horses and pedestrians. A case is on record of a 
horse being thrown to the ground by bridging the two rails of a 
track. Actual test showed a difference of potential between the 
rails of 8o volts. The question of the intervals at which cross 
bonds should be installed is a pertinent one on which further dis- 
cussion is to follow. 

Rail Joints and Bonds 19 

(6) Special- Work Bonding. — The problem of bonding special 
work offers some features not present on straight track. The joints 
are difficult to maintain, pounding due to traffic is excessive, fre- 
quently hard steel is used to which certain types of bonds can not 
well be applied, the sharp angles in frogs exclude the use of certain 
bonding tools, and the replacement of steel is more frequent than 
on straight tracks. Owing to these complications it has been found 
difficult to maintain good bonding through special work, and as a 
result it has become standard practice to bond around all such sec- 
tions with heavy conductors. These are relied upon to carry the 
main current, and light bonding is usually installed to take care of 
the current originating in the section thus shunted out. Cables 
employed for this purpose can not be left exposed in unguarded 
regions as they offer particular temptation to copper thieves. 
Even when buried in. the earth they are frequently dug up and 
removed by junk vendors. 


With the advent of heavy electric traffic in paved city streets 
the bond and joint problem became acute. Both were found 
difficult to maintain and the repair of either entailed considerable 
expense in removing and replacing pavements. The rail joint 
was recognized as a weak link and the limiting factor in the life of 
their tracks by the operators ; and when it was proposed to obliter- 
ate it by welding the joint, and was demonstrated to be a practical 
possibility, the idea was eagerly incorporated by many companies. 
Although the welded joint, like the rail bond, has had a somewhat 
checkered history and has met with many reverses and failures, it 
has grown in use until at the present time it is a standard of con- 
struction in some form or another in nearly every large city in the 
United States. The several types of welded and special joints 
which are now in service will be briefly described here, leaving a 
fuller discussion to the operating companies whose testimony is to 
follow later. • 

(a) Cast Weld. — This term is used in reference to an early type 
of weld, sometimes called the Falk joint, produced by pouring 
molten cast iron into a mold around the ends of the abutting rails, 
the latter having been cleaned by a sand blast or some other means. 

20 Technologic Papers of the Bureau of Standards 

In many cases it was not a weld at all owing to the difficulty in 
preheating the cast iron sufficiently to melt the steel of the rails. 
It was a good mechanical joint, however, and frequently served 
admirably the functions of both bond and joint plates. The many 
failures that occurred in this and other types of welds have been 
attributed to numerous causes. Whether the heating of the rail 
was sufficient to change its properties, and thereby result in exces- 
sive wear and breaks, has occasioned a great deal of discussion and 
is still a mooted question. Expansion and contraction has un- 
doubtedly been responsible for failures in some cases, particularly 
where rails were welded in hot weather or laid in pavements having 
poor binding qualities. The cast weld is still doing duty in a 
number of localities, but in competition with other types, which 
are accomplishing the desired results in a more satisfactory man- 
ner, its installation has been practically abandoned. 

(b) Thermite-Welded Joint. — This is essentially a cast weld, 
but as the temperature attained by the reaction of aluminum and 
iron oxide is far in excess of that produced in the ordinary cast 
weld, the ends of the steel rails are melted down and an oblitera- 
tion of the joint actually results. This method found early and 
wide application on experimental scales owing to the comparatively 
simple manner of obtaining a complete weld of the joint, but 
later it suffered a setback owing to the large number of joints 
which were found to cup after bearing heavy traffic for some 
time. In order to properly make the weld it was necessary to 
leave a space of about three-quarters of an inch between the rail 
ends and fill in with the molten metal. As it was practically impos- 
sible to duplicate the physical properties of the rail a soft spot was 
left which eventually gave rise to cupping. Recent improvements 
"have eliminated this defect, however, and the joint is again 
finding favor with the railways. It requires much less metal 
than the old cast weld, and has a high conductance. It is better 
adapted for new than for repair work, as the process is difficult 
to carry out under traffic conditions. A completed weld is 
illustrated in Fig. 9. 

Thermite welds have also been used in connection with mechan- 
ical joints. A shoe of thermite steel is poured around the base 
of a bolted or riveted rail joint, thereby welding the base of the 

Bureau of Standards Technologic Paper No. 62 

Fig. io. — Electrically welded joint with head support 

Fig. ii. — Arc welded joint 

Fig. 12. — Arc welded joint with head support 

Rail Joints and Bonds 21 

rail but in no wise changing the properties of the head or running 
part. The joint needs no other bond and is mechanically good. 
It has met with marked success in Baltimore and Cleveland, where 
it is known as the " Clark joint." 

(c) Electrically- Welded Joint. — The electrically-welded 
joint was introduced over 15 years ago and has found wider applica- 
tion than any of the other modern types. It consists of heavy bars 
or plates from 2 to 3 feet in length spot welded to the web of the rail 
by the use of an electric current. The process requires a heavy and 
expensive plant and is usually carried out by contract on a com- 
paratively large scale. For this reason it is not well suited to 
installations on small systems. It is well adapted to the re- 
claiming of old track as well as for new work and has been applied 
on open T-rail construction where expansion joints are installed 
at intervals to provide for expansion and contraction. Fig. 10 
shows a section through the center of an electrically- welded joint 
with head supports. 

Failures of this type of weld have in general been confined to 
fracturing of the rail at the end of the welded bar and are no doubt 
the result of strains introduced into the web of the rail from the 
localized heating. Such breaks have been more prevalent in 
old than in new rails. On new work with rails having no bolt 
holes the failures have been reduced to a minimum, which is not 
considered serious. 

(d) Arc-Welded Joint. — This term will be employed in 
reference to joints in which the plates have been welded to the 
rail by means of an electric arc. The method is comparatively 
modern and is just emerging from the experimental stage. The 
features which appeal to the operating companies are ease and 
simplicity of application, low cost, and high conductance. In the 
most recent types the rails are not heated to an injurious degree. 

This method of welding has also been applied to the welding of 
rails to steel ties and also to the welding of bolted and riveted 
joint plates to the base of the rail. This latter process provides 
for a joint of high conductance, strengthens it mechanically, and 
eliminates any possible danger from heating the web and head of 
rails. The recent adoption of this process for welding joint plates 
by numerous companies after a trial installation is strong evi- 

22 Technologic Papers of the Bureau of Standards 

dence of its utility for this purpose. Two improved types of 
welded joints are shown in Figs, n and 12. 

(e) The Nichols Composite Joint. — This joint consists of two 
plates which tit the web of the rail and are riveted snugly to it 
but which do not come in contact with the fishing surfaces. The 
spaces thus left around the head and foot of the rail are filled with 
molten zinc, which expands upon solidifying and enters into all 
of the irregularities of the rail surface. If the rail and plates 
are properly cleaned before pouring the zinc, a good electrical 
contact is obtained and a joint of high conductance is the result. 
The process is rather expensive, and is warranted, therefore, only 
on lines bearing heavy traffic. The fact that it has been a standard 
of construction in the city of Philadelphia for a number of years 
is an indication of its practicability. 

(/) Mechanical Joints. — In recent years a number of com- 
panies have gone to some pains and expense in the construction 
of bolted and riveted joints designed to eliminate the effects of 
expansion and contraction, which in the ordinary bolted joint 
often results in looseness and ultimate failure. Where bolts are 
used, the holes are drilled undersized and carefully reamed out to 
give a driving fit. Machine bolts with square heads are used and 
an accurate shop fit is obtained in the field. Rivets are some- 
times substituted for bolts. In some cases the plates are drilled 
short and heated to make the holes meet. When they shrink, the 
ends of the rails are drawn together with a great force and trouble 
from cupping is thereby largely eliminated. Such mechanical 
joints as here described, when carefully bonded, possess the best 
features of the welded joint and are free from some of the latter' s 
defects. It is obvious that they can not be used in other than 
paved streets unless some provision for expansion and contraction 
is made. 



Before asking for information or opinions from any of the 
operating companies, a thorough study of the bond and rail joint 
problem was made by a careful reference to all trade literature 

Rail Joints and Bonds 23 

available, by correspondence with manufacturers, and review of 
the technical literature of the subject. 

After obtaining a fair knowledge of the problems involved and 
the present practices employed, a circular letter and list of ques- 
tions were prepared for distribution, among the operating com- 
panies. The letter called attention to the lack of uniformity 
among the operating companies in their bonding practices and 
the need for cooperation in an effort to better solve the difficulties 
connected with the problem. It asked for a free discussion of the 
subject and for individual opinions rather than a reply confined 
to the answering of definite questions. This letter, together with 
a copy of the questions, was mailed to a list of 130 operating com- 
panies selected from every State of the Union, made up for the 
most part from the larger companies, although a number of com- 
panies operating railways in cities of from 1 5 000 to 40 000 popu- 
lation were included. 

Following is a list of the questions submitted: 


i. What types and sizes of rail bonds have you used? State whether soldered, 
brazed, welded, compressed, *or expanded terminal bonds; whether solid, stranded, 
or ribbon; whether concealed or exposed; to what part of the rail attached; number 
and length of bonds per joint. 

2. On what type of construction has each type of bond been used? Give weight 
of rail, length of rail, type of rail, type of splice bars, type of roadbed, kind of pave- 
ment adjacent to rails, character of traffic. 

3. About how many of each type and size of bond have you in operation at the 
present time? 

4. Do you inspect your bonds regularly? If so, how, and how frequently? 

5. Do you have a definite criterion for determining when a bond requires replacing? 
If so, what is it? 

6. About what is the average life of each type of bond? Give the percentage of 
failure for each year after installation. 

7. At what intervals do you cross bond the two rails of each single track; also the 
two tracks on double-track construction? 

8. On what basis do you determine the size of cross bonds to be used? 
9: Do you bond around all special work? 

10. What type of bonds do you use for cross bonding and special-work bonding? 

11. To what extent have you been troubled with theft of different types of bonds? 

12. What grade of labor do you employ for the installation of different types of 
bonds, skilled or unskilled? 

13. What special tools and machinery do you employ in the installation of dif- 
ferent types of new bonds; also, in the repair of bonds? 

14. Do you use a lubricant in drilling for compressed or expanded terminal bonds? 

24 Technologic Papers of the Bureau of Standards 

15. Are your drills machine or hand ground? 

16. What is the complete cost of installing each type of new bond when installed 
in large numbers; also, of renewing a bond? State in detail how costs are com- 
puted. (The total cost of installing a bond should include the cost of the bond, 
the cost of the tools, the interest and depreciation on the tools and machinery, sup- 
plies, electric energy if used, labor,, and paving repairs.) 

1 7 . What is the most prevalent cause of failure of each type of bond ? ' 

iS. What is the average resistance in terms of rail length, or in ohms, of each type 
of bonded joint when new and when old? 


1. What types of rail joints, and how many of each type, have you in operation at 
the present time ? 

2. On what type of construction has each type of joint been used? Give weight 
of rail, length of rail, type of rail, type of roadbed, kind of pavement adjacent to rails, 
and character of traffic. 

3. What is the average efficiency of each type of joint, i. e., its electrical resistance 
compared to an equal length of rail ? Give length of joint used in testing. 

4. About what is the average life of each type of joint? Give the percentage of 
failures for each year after installation. 

5. What is the total cost of each type of joint on new work? What is the cost of 
replacing a joint which has failed? (See note to question 16 under "Rail bonds. ") 

6. What is the most prevalent cause of failure in each type of joint? 

7. What are the maximum and minimum temperatures to which your welded 
joints are subjected ? 

8. With what types of joints and under what conditions do you use expansion joints 
in your rails? 

In answer to the above inquiries, 42 companies, or 32 per cent 
of those addressed, submitted replies. This percentage, although 
not large, was considered as satisfactory and encouraging. The 
mileage represented in the replies was about 8600 miles of single 
track and represents nearly 20 per cent of the mileage in the 
United States. The constant demand upon public-service cor- 
porations from outside sources for statistics and information has 
assumed such proportions that it is little wonder that a large 
number of them have been disregarded altogether. Information 
is sought by engineering committees, colleges, State and city 
commissions, Government bureaus, and private engineers, and 
unless accompanied by an authoritative request such communi- 
cations frequently are pigeonholed or find their way into the 
wastebasket. The generous number and the nature of the re- 
plies received, as well as the apparent willingness on the part of 
the companies to cooperate with the Bureau in its investigation, 
were complete vindication for the attempted study. The attitude 

Rail Joints and Bonds 25 

of the companies toward the matter is illustrated by the follow- 
ing statement from L,. P. Crecelius, superintendent of power for 
the Cleveland Railway Co. and president of the engineering asso- 
ciation of the A. E. R. A. for the year 1 914-15. 

We agree in the suggestion that a study of the question of rail bonds is very 
important. Examination of the practice in regard to rail bonds reveals a wide 
difference in opinion as regards both the type and character of rail bonds and the 
method of applying them. We consider the subject not only important, but timely, 
and take pleasure in replying to your circular letter requesting cooperation. 

With but a single exception the answers were submitted in 
good form, and the majority of them showed considerable labor 
in their compilation. A few replies showed the marks of a hasty 
and cursory survey of the subject, but these constituted only a 
small minority. 

Below is given a list of the companies and their equivalent 
single-track mileage from whom replies to the circular inquiry 
were received. They are listed and numbered in the order in 
which the replies were received, and hereafter, for the sake of 
brevity, will be referred to by number rather than name. The 
mileage here given was taken from the McGraw Electrical Direc- 
tory of August, 191 5: 


i. Union Electric Co., Dubuque, Iowa 21 

2. Ohio Electric Railway Co., Springfield, Ohio 617 

3. Indianapolis Traction & Terminal Co 158 

4. Milwaukee Electric Railway & Light Co 403 

. 5. Dallas Consolidated Electric Street Railway Co 50 

6. United Railroads of San Francisco 280 

7. Helena Light & Railway Co .• 24 

8. Scranton Railway Co 100 

9. Puget Sound Traction, Light & Power Co., Seattle, Wash 499 

10. Connecticut Co., New Haven, Conn 171 

11. United Railways & Electric Co. of Baltimore, Md 405 

12. Houston Electric Co 75 

13. Spokane, Portland & Seattle Railway Co 240 

14. Chattanooga Railway & Light Co 61 

15. St. Joseph Railway, Light, Heat & Power Co 49 

16. Harrisburg Railways Co 72 

17. Omaha & Council Bluffs Street Railway Co 140 

18. Metropolitan Street Railway Co., Kansas City, Mo 265 

19. Memphis Street Railway Co 124 

20. Tacoma Railway & Power Co 104 

21. Everett Railway, Light & Water Co 17 

22. Butte Electric Railway Co 35 

20 Technologic Papers of the Bureau of Standards 


23. Terre Haute, Indianapolis & Eastern Traction Co 101 

24. Louisville Railway Co 150 

2$. San Francisco-Oakland Terminal Railways 256 

26. Portland Railway, Light & Power Co., Portland, Oreg 299 

a 7 . Minneapolis Street Railway Co 435 

28. Pacific Electric Railway Co. , Los Angeles, Cal i>°57 

29. Los Angeles Railway '. 383 

30. Omaha, Lincoln & Beatrice Railway Co 56 

3 1 . Birmingham Railway, Light & Power Co , 154 

32. United Railways Co. , St. Louis, Mo 442 

2S- Interborough Rapid Transit Co., New York, N. Y. 167 

34. Rhode Island Co. , Providence, R. 1 399 

35. New York Central & Hudson River Railroad Co 254 

36. United Traction Co., Albany, N. Y 103 

37. Cleveland Railway Co 344 

38. Peoples Railway Co., Dayton, Ohio 31 

39. Pennsylvania Railroad Co 99 

40. Cleveland Southwestern & Columbus Railway Co., Cleveland, Ohio 225 

41. Philadelphia Western Railway Co., Upper Darby, Pa. 50 

42. Washington, Baltimore & Annapolis Electric Railroad Co 100 

Realizing that the information obtained from formal replies to 
a set of questions could be greatly strengthened and its value 
enhanced by personal interviews with the engineers of a number 
of companies a representative was placed in the field for that 
purpose. His itinerary included Harrisburg, Baltimore, Phila- 
delphia, New York, Boston, Buffalo, Cleveland, Cincinnati, 
Indianapolis, Louisville, Richmond, and about 20 other interme- 
diate cities and towns. During his five weeks' tour he interviewed 
upward of 50 electric railway engineers and visited a number of 
manufacturing establishments. 

The information obtained on this tour, while not abounding in 
figures and percentages, forms a valuable supplement to the 
carefully prepared and conservative statements and tabulations 
contained in the formal answers and will be included when possible 
in the summations which are to follow. 

Owing to the large number of questions asked and to the nature 
of the answers the material does not well lend itself to a con- 
densed tabulation. Little could be gained by such a summary 
owing to the indefiniteness of the replies and the fact that they 
consist largely of estimates and opinions. 

A more satisfactory presentation of the data and the one which 
will be employed is a tabulation of figures, facts, and opinions 

Rail Joints and Bonds 27 

under the several questions asked, to be followed by an analysis 
of all available data, with recommendations, under the various 
types of bonds, joints, and methods in use. 

As the amount of material collected is so large as to preclude it 
from publication in full, summaries will be made where prac- 
ticable. On points of particular interest and on which a lively 
difference of opinion exists or on which opinion has not yet 
crystallized the information as received will be given in full. 

While there is, of course, no thought or desire of injuring or 
depreciating any product of any manufacturer, a frank and 
impartial discussion of all materials and methods is essential to 
the best results of the investigation. Such discussion, it is believed, 
will not injure any standard product or method, but, on the other 
hand, will benefit the manufacturers in helping the companies to 
standardize on certain products, thereby relieving the factories 
from continuing the manufacture of the great variety of types 
that are now demanded by the trade. 



(a) Questions 1,2, and 3. Number and Types of Bonds. — 
These questions ask for the sizes and types of bonds which have 
been used, for the type of construction on which each type of 
bond has been used, and for the number of each type in use at 
the present time. 

In answer to the first question a majority of the companies 
enumerated a number of types of bonds as having been used in 
the past and stated that most of the early types had been replaced 
by modern bonds. The following answer from the Louisville 
Railway Co. is given to show the great variety of bonds which 
have been employed since the early days of electric railway engi- 
neering and serves admirably as an historical sketch of the devel- 
opment of the practice as well as an example of the experimental 
efforts which the roads have resorted to in their attempt to 
suitably bond their tracks. 


A. No. 6 bare tinned copper wire bent around a -Mi-inch copper rivet in each end 
and placed»outside of splice. On some occasions soldered to a continuous wire of same 
size running along the track; used about 1889. 

aS Technologic Papers of the Bureau of Standards 

B. } o Chicago bond, solid type, used outside of splice bar, rivet driven in from 
back; used 1S94, 1895, and 1896. 

C. 40 solid crown bond adopted in 1897; used outside of splice. Steel drift pin 
driven in from front. 

D. In same year tried on 2 miles of track the Bryan and brass washer, bolted type, 
using two pieces of 1/0 trolley or two pieces of 4/0 insulated. 

E. Used in 1899 and 1900, 300 4-inch protected bonds (4/0 flexible), which were put 
on with screw compressor — T rail. They were too short to allow for contraction and 
were replaced a few years later by 12-inch crown bonds. 

F. 300 000 c. m., 500 000 c. m., and 1 000 000 c. m., cable "Buffalo" bonds, cast- 
brass terminals threaded for nut, bolted, soldered, and tinned on web of 9-inch girder 
rail; 1899 to 1901. 

G. SX-inch to 12-inch flexible (4/0) crown riveted bonds used under Atlas and con- 
tinuous joints on T-rail track, interurban and park line 1902 to 1907. 

H. Adoption of same type on 9-inch girder rail. 

I. 68-inch and 6i^-inch solid (4/0) crown riveted cross bonds. 

J. Previous to this cross bonding had been done by soldering continuous wire to 

K. Adoption of solid terminal, compressor type, except where compressor could 
not be used around special work, short bonds flexible, long ones solid; 1908. 

L. Installation in 19 10 of exposed brazed ribbon bonds outside head of rail on 40 
miles 70-pound T track; interurban. Found large number of these bonds broken in 
succeeding years, partly due to loose and low joints. 

M. A trial in 1910, on one side of 6}4 miles of an interurban line 1000 of the American 
Steel & Wire Co.'s twin terminal type; 75 per cent stolen in 1913-14 and the rest 

N. Returned in 1913-14 to 4/0 crown flexible compressor type under continuous 
joints on all open track. 

O. Cross bonding on 70-pound T rail in country by pieces of 70-pound rail turned 
upside down, welded to base of running rail every 1000 feet, during 1914. 

P. Welding 300 or more pairs of splice bars (1914) on 7-inch and 9-inch girder rail, 
city construction, using Indianapolis arc welder. 

Q. One bond to each joint is used on 60-pound or 70-pound T rail or old light girder 
rails. One, two, or three bonds each joint on 7-inch or 9-inch girder rail in the city, 
according to headway of cars. 

R. All bonds attached to web of rail except in items L and M, which were attached 
to rail heads. 

Expanded bonds, A, B, C, D, G, H, I. 

Compressed bonds, E and K. 

Soldered bonds, F. 

Standard type bonds, G, H, K, M, N. 

Question 2, as to the type of construction on which each type 
of bond had been used, was not asked so much with the idea of 
determining what bond is best suited to a given type of con- 
struction as to determine, if possible, the causes of reported 
bond failures. In a few instances the answers lend themselves 

Rail Joints and Bonds 29 

to such interpretation and will be employed in the discussion of 
that subject. 

In answer to the third question 36 of the 42 companies gave 
some figures as to the number of bonds in use at the present time. 
In some cases their figures were not complete, but were given for 
only one or more types of bonds in use or for bonds installed since 
a given date. The number of bonds reported, therefore, is only 
a fraction of the total number represented by the 36 companies 
answering this question, but nevertheless undoubtedly give a 
fair indication of the relative number of each type that has been 
installed in recent years throughout the country. 

The 2 305 300 bonds reported are classified as follows: 

Compressed terminal: 

Ribbon, concealed 214 200 

Stranded, concealed 338 100 

Stranded, exposed 176 300 

Not described 201 000 

Total 929 600 

Pin terminal: 

Stranded, concealed 241 100 

Stranded, exposed 63 200 

Not described 58 500 

Total 362 800 

Electric weld: 

Exposed 392 600 

Concealed 27 400 

Not described 300 000 


720 000 

Soldered, all kinds . 140 800 

Twin terminal 113 700 

Oxy-acetylene, welded 15 000 

Channel pin 20 000 

Plastic alloy 3 400 

Total 292 900 

Grand total 2 305 300 

These figures show that among the companies reporting the use 
of compressed-terminal bonds is in excess of that of any other 
type, with the electric-weld bond second. 

30 Technologic Papers of the Bureau of Standards 

For concealed application tne stranded bond is a favorite over 
the ribbon type, the percentages being 61 and 39, respectively, 
in the compressed-terminal type, while no concealed-ribbon bonds 
of the pin-terminal type were reported. 

The largest number of bonds reported was 315 000 by Com- 
pany 28. Of these 300 000 are electric -weld and include both 
head and web types, the relative numbers not being given. 

The second largest number of bonds reported was 243 000 by- 
Company 35. Of these 81 000 are soldered bonds on third rail 
and 162 000 pin-terminal, stranded bonds. Company 6 reports 
the use of 199 500 bonds, 196 000 of which are said to be com- 
pressed terminal. They no doubt include all varieties of that 
type and are listed as "not described." Of the 720 000 electric- 
weld bonds in use 300 000 were reported by Company 28 and 
160 000 by Company 2, and of the 113 700 twin- terminal bonds 
60 000 were reported by Company 1 1 . The 1 5 000 oxy-acetylene- 
welded bonds are used by Company 27. The two million bonds 
and more reported are sufficient to single bond 30-foot rails on 
6550 miles of single track. As this is about 75 per cent of the 
mileage represented by the answering companies, it is safe to 
assume that a large percentage of bonds in use were reported. 

(b) Question 4. Inspection op Bonds. — Forty-two answers 
to this question are briefly summarized as follows: Fifteen com- 
panies do not inspect bonds regularly ; 6 companies test bonds by 
inspections and melting snow only; 2 companies test bonds every 
3 months; 10 companies test bonds every 6 months; 12 companies 
test bonds every 12 months; 1 company tests bonds every 18 
months; 3 companies test when bonds are apparently in bad 
condition; 4 companies employ an autographic-recording bond- 
testing car; 23 companies employ some kind of portable bond 
tester; and 1 company tests exposed soldered bonds with a hammer. 

These figures indicate that all degrees of inspection and testing 
are employed from the observance of joints around which snow is 
melting to frequent tests with accurate instruments. While the 
majority of the companies aim at regular inspection and testing, 
numerous interviews revealed the fact that such aims are not 
always realized. Testing is frequently done by men who have 
other regular duties to perform and who give their time at irregu- 

Rail Joints and Bonds 31 

lar intervals to this work, and this time may not always be sufficient 
to comply with the standards set by the company. Moreover, in 
times of financial stress this feature of maintenance is frequently 
considered as unnecessary and is consequently slighted. Even 
were it carried out, the funds for following up the replacement of 
poor bonds are not always available. 

Several companies state that the condition of exposed bonds is 
determined by inspection only, instrument tests being confined 
to concealed bonds. While bonds can be quickly inspected by 
trackwalkers this does not always give assurance of their good 
condition. Mechanically applied and soldered bonds frequently 
develop high resistances when no indication of such is apparent 
from inspection. A hammer blow gives a better indication of 
conditions, but of course is not always infallible. 

It should be pointed out that failure to test bonds regularly and 
systematically is not always an indication of carelessness or 
unconcern on the part of an operating company. A test of bonds 
is of little value unless followed up by repair. It frequently 
happens that a systematic repair of bonds, as a bonding proposi- 
tion only, is not justifiable owing to the fact that a general rehabili- 
tation of the system may be in progress or in contemplation, 
which in the course of a very few years would weld or otherwise 
repair all joints. Under such circumstances a company might not 
be justified in repairing any but the very worst joints, and these 
as a rule can be detected either by inspection or by melting snow. 
This condition while not often existing over an entire system is 
frequently met with in limited areas or regions, where it becomes 
necessary to maintain an old track for from one to three years to 
await the city's order for change of grade or for repaving. A 
general repair of bonds under such circumstances might not be 
advisable, and again only open joints would receive attention. 

(c) Question 5 . Criterion for Replacement. — Upon this point 
widely varying standards appear to be in practice. Twenty-six 
companies of the 42 have a definite resistance below which they 
aim to maintain all joints, the resistance being defined in terms 
of equivalent length of adjacent rail. Other roads bond only 
when joints are open or bonds are broken, or where snow is found 

2,2 Technologic Papers of the Bureau of Standards 

to be melted. The summary following is compiled from answers 
submitted to question 5. 


Criterion for Bond Replacements 

Number of companies 

Criterion for re- 
placement (3 feet 
of joint equal 
to feet of rail 

Number of companies 

Criterion for re- 
placement (3 feet 
of joint equal 
to feet of rail 





























° Open bond or melted snow, or when 50 per cent of strands are broken. 

6 No answer. 

Joints showing a resistance of 4 feet of rail or better are, as a 
rule, bonded with two or more bonds, and are confined for the 
most part to heavy traction lines such as exist in New York City. 
The average value of the figures for city streets is 8.3 feet. 

(d) Question 6. Average Life of Bonds. — The answers to 
this question are so vague and inconsistent and the failure of 
bonds are affected by so many conditions and circumstances that 
it is difficult to draw definite conclusions from them. The fact 
that the life of a bond depends in a large measure upon the life 
of the joint was brought out by a large number of the answers. 
Few figures and statistics giving percentages of failures and the 
life of various bond installations were submitted. The question is 
one of such great importance and the answers of so much interest 
that a number of them are here quoted. 

Company 3. — The brazed type of bond shows a failure of 2 per cent, which is caused 
by loose rail joints, and is not charged against the bond. The compressed type shows 
no failures chargeable to bonds. * * * Where the failures of compressed type of 
bonds may appear unreasonable, we have no record of this type of bond failing in the 
past 10 years other than due to rail breaks or where joint becomes so loose that it 

Rail Joints and Bonds 33 

would gradually break the strands and separate the bond. All of our bond terminals 
are carefully tinned before compressing in the rail. We take particular pains in 
grinding our drills so as to make a smooth, clean hole in the rail, and also keep the 
bond compressor in good condition so as to get an even compression on the bond 
terminal. We have kept this practice up for 10 years and find that it is giving excel- 
lent results, and we now figure the life of a bond longer than the life of the rail. 

This company uses stranded bonds and continuous splice bars. 

Company 4. — We have approximately 34 000 soldered leaf bonds on our open track. 
Of these about 3500 were replaced in 1912, about 2200 in 1913, and about 5100 in 1914. 
Our records do not show the percentage failure by years after installation. 

Company 5. — The life of a bond we have found to be determined entirely by the 
life of the joint, i. e., so long as the bolts remain tight and the joint in good service, 
just so long will the bond be good. In nearly all cases where defective bonds have 
been removed it has been found that the joint itself was loose and by its vibration 
the bond would become detached from the rail or at least become loosened. 

This company employs stranded pin-terminal bonds. 

Company 6. — The life of a bond depends almost entirely upon the rigidity with 
which the rail joint is maintained. With an absolutely unmovable practically homo- 
geneous rail joint one bond would last about as long as another and all would last until 
the rail was worn out. Data about "average life" of bonds would be misleading. 
Such data has not been kept. Our experience has taught us to prefer short expanded 
terminal, concealed, ribbon bonds where permissible. 

Company 7 uses twin-terminal soldered bonds and renews 
them when they show a resistance greater than 15 feet of rail. 
In answer to question 6 they say : 

Company 7. — The average life of the bond is about 6 years. Nearly all of the 
failures were either due to poor workmanship in putting them on, or to the loosening 
of the splice bars which allowed a movement of the rails to break the strands of the 

Company 8. — We have no data on which to base the probable life of the different 
types of bonds. It has been our experience that the mechanical condition of the 
joint fails before the bond. Consequently the failure of the bond is then due to the 
poor mechanical condition of the joint. Should the joint be maintained in good 
mechanical condition and the bond is properly installed, cases have been noted 
where the bond did not show any depreciation after 8 to 10 years of service. These 
instances are rare, for the reason that the mechanical condition of the joint usually 
requires attention before that period, and the bond is also replaced. 

From our experience the life of the exposed bond is greater than that of the con- 
cealed type, probably due to the fact that the mechanical condition of the track is 
usually better. 

Company 10. — The life of the bonds, we find, vary from 3 to 12 years. The per- 
centage of failures can not be stated at this time. 

Company ii, — Twin-terminal and compressed-terminal bonds usually last as long 
as the track, the percentage of failure is very small. The percentage of failure of 
soldered bonds is extremely high; we have no definite figures. 

Company 12. — The average life of these bonds is about 6 years. 
150211° -19 3 


Technologic Papers of the Bureau of Standards 

This company uses stranded concealed pin-terminal bonds. 

Company 13. — Our soldered bonds, after 7 years, are practically all off, and 50 per 
cent of them were off in half that time. Our experience with these bonds has been 
that the few bonds which we were able to get on properly gave excellent results, but 
on the average it was impossible to get these bonds so soldered to the rail that they 
would stand under heavy traffic. 

At the present time our interurban lines outside of the city, consisting of about 
150 miles, are virtually all bonded with the welded bonds. 

We aim to make an inspection of our bonds at least once a year. Our trouble with 
the welded bonds has been the breaking of the leaves of copper just below the lug 
which is welded to the rail. This is doubtless due to the weakening of the metal 
through heating at the time the weld is made. Vibration and the deflection of the 
joint tend to work the leaves till eventually they snap off close to the lug. We have 
endeavored to overcome this by having the bond people furnish us with a welded 
bond with the terminal in the shape of a T, so as to overcome any bending of the 
leaves by deflection at the joint. We have not had these experimental types of bonds 
on long enough to tell whether the cure is going to be effective. 

In the case of the pin-terminal bond, the only trouble experienced has been the 
loosening of the terminal by reason of incomplete expansion when driving the pin. 
These bonds have to be very carefully put on or they will work loose under the plates. 
The compressed-terminal bonds have given us better satisfaction in this respect. 

The following table, giving percentage of failures for various 
types of bonds, is extracted from the report of Company 14: 

Company 14. Bond Failures 

Kind of bond 

Between rails 

tage of 






































in miles 



compressed . 
compressed . 
compressed . 
compressed . 
compressed . 
compressed . 

Brick. ... 

Wood block. 
Macadam. . . 





Macadam. . . 




Macadam. . . 




Rail Joints and Bonds 


Company 16. — The life of a bond depends almost entirely upon the type of 
track construction. In first-class construction, such as we use in paved streets, we 
never have a broken bond, and the depreciation is confined to a slight corrosion of 
the strands, and in a few cases a corrosion of the terminals. The latter only occurs 
where the bond has not been properly applied. 

In open-ballast construction, expansion and contraction of the rail must be allowed 
for in the joint. This, with defects in track construction, such as loose bolts, poor 
foundation, etc., causes the strands on the bonds to break. On suburban lines the 
record for the year 1914 is as follows: 

Number of joints tested 


Type of joint 

of rail 

Age of 


Per cent 











56 and 65 

70 and 85 













Company 17. — Up to the present time we have noticed no failure of the brazed 

Company 19.— Bonds will last indefinitely if rail joints are kept in good repair. 

Company 22. — No record. If the bond is correctly installed and the rail joint is 
perfect, I see no reason why the bond should not last as long or longer than the rail. 

Company 23. — I am unable to state what is the average life of each type of bond, 
but in my opinion the short bond attached to the head of the rail will, as a rule, last 
3 to 6 years and the larger compressed-terminal bond around the splice will have an 
average life of approximately 12 years. The life of all bonds depends entirely upon 
the track conditions. If it were possible to keep the joints absolutely tight under all 
conditions it is quite possible that the life of the bond, especially of the longer type 
around the splice, would be equal to the life of the rail. 

Company 24. — Brazed bonds, 7>£ years; compressed bonds, 15 years. 

Company 25 submitted a table showing the cause and rate of 
failure of about 10 different sizes and types of bonds. All types 
show a high rate of failure with the exception of the compressed 
terminal, installed with mercury alloy, which shows no deprecia- 
tion. This type, however, is the only one used on paved streets 
with concrete roadbed, and this type of construction no doubt 
has more to do with their good performance than the type of bond 

Company 26.— Can not state average life of bonds. The failure of compressed-ter- 
minal bonds has been very light, probably not 2 per cent in six years; on twin-terminal 
bonds would estimate 5 per cent failure in five years. The soldered bonds have been 
unsatisfactory. We have found it impossible to maintain them in good condition. 

36 Technologic Papers of the Bureau of Standards 

Company 27. — We find that all pin-terminal bonds, expanded in holes in the rails, 
after a comparatively short life of low resistance begin to blacken at contact, with 
resulting high resistance within one year after installation. We find that with soldered 
terminals the solder disintegrates either from vibration or electrolytic action, and that 
within one or two years a high resistance results. This is also true where heavy lugs 
are soldered onto the rails for the purpose of connecting cables. 

Company 29. — Ten-inch compressed-terminal ribbon bonds have been in service 
over 10 years; no trouble from breakage. All concealed bonds have an average life 
of from 5 to 15 years, depending upon maintenance of plate bolts from loosening up. 

Company 3 1 . — The life of a bond depends more on the condition of the rail joint than 
any other feature. We have never kept any records that would give us actual life of 
bonds under various conditions, but we believe that in our recent concrete paving 
construction, where we desire rail joints tight and pay special attention to their instal- 
lation, the life of the bonds will equal the life of the rail and paving, which will be 
from 10 to 25 years, depending on the quantity of traffic. We believe further that 
on some of our old track of 56 and 60 pound rail, either in open work or chert paving, 
where the rail joints are very bad, the life of the bond is probably only a few months, 
or at least one year. 

Company 3$. — On account of the fact that the life of the bond in a track rail 
is contingent on the life of the rail, and as the life of this track rail is comparatively 
short, due to the high frequency of service and the heavy car mileage, an estimate of 
the average life of the bond would not be reliable and might be misleading. 

Regarding the life of bonds on the contact rail, would advise that except in isolated 
instances the bonds show no appreciable deterioration after a life of 12 years. 

Company 34. — In 1899 about 6 miles of 70-pound, 60-foot rail, single track, was 
laid in country road, dirt ballast, bonded with two 36-inch, 4/0 stranded bonds 
with expanded head. Tested 8 years after installation, showed no defective joints. 
Tested 15 years after installation, showed about 4 per cent defective joints. 

In 1901 about 30 miles of single track was laid in country roads with 70-pound T rail, 
60 feet long, bonded with one 8-inch and one 17-inch, 4/0 concealed bond with 
expanded head, around each joint. These bonds were tested in 1914 and showed 
about 4 per cent defective. 

In 1904 about 21 miles of single track was laid with 9-inch girder rail, 60 feet long, 
98 pounds per yard, gravel ballast, and block and macadam paving. Bonded with two 
12-inch ribbon, 4/0 bonds, with compressed heads. Tested in 1907, no defective 
joints; tested in 19 14, showed about 5 per cent defective joints. 

In 1913 we rebonded somewhere over 5000 joints on 60 and 70 pound T rail, 30 and 
60 feet long, with 9-inch 300 000 c. m. ribbon bonds welded to ball of rail. Tested in 
19 14, no defective joints. 

Company 35. — Average life of bond not yet determined. Percentage of failures: 
1909, 0.15 per cent; 1910, 0.069 per cent; 1911, 0.082 per cent; 1912, 0.064 per cent; 
1913, 0.047 P er cent; 1914, 0.098 per cent. 

This company uses pin-terminal stranded bonds. 

Company 36. — I can not give you definite information as to the average life. In 
1913, 0.86 per cent of bonds were replaced; 1914, approximately 0.1 per cent. 

This company uses pin-terminal stranded bonds of various 

Rail Joints and Bonds 37 

Company 37. — Life of bonds has been found to depend very largely upon the type 
used and its exposure to vehicle traffic, tampering, and theft. A loss of 2 per cent as 
failures in brazing rail bonds to the rail we have found to be an average figure and as 
representing the failures due to defective brazing, which, however, are checked up and 
corrected as the work progresses and such defects are made good. The brazes show 
a remarkably effective life, and failures seem to take place in the breaking off of the 
strands next to the terminal. Our experience for the last few years indicates that a 
thousand bonds per year become defective because of theft, mutilation by vehicles, 
and breaking off of strands at the terminal, or based upon the total number of bonds 
we have in service this is approximately 1 . 5 per cent loss per annum. We have bonds 
in service which were brazed to the rails 10 years ago. 

Company 38. — Over 30 miles of track which has been installed from the year 1894 
to the present time the percentage of failure you will find as follows: Special work — 
open, 235; defective, 81; main track — open, 181; defective, 129; giving figure of 
merit, 87.8. 

These figures were derived from autographic records recently 
made by bond-testing car. 

Company 39. — The average life of each type of bond varies in our installation. 
The exposed type of bond is used in tunnels and station area, and in three years we 
have had to renew 65 bonds, or about 0.4 of 1 per cent of the total number installed in 
these sections. 

Of the concealed type we have had to renew 1^00, or about 5 per cent of the total 
number of bonds installed. The large renewal is confined to one section, which is on 
built-up ground where high speed with heavy traffic loosens the joints. We have 
also found that in some cases the splice bars squeeze the strands of the bonds and they 
eventually break off. This of course is not due to defects in design and manufacture of 
bonds. The percentage of failures for the first year was less than 1 per cent, and for 
the past two years was a little over 0.1 per cent, and for the last year was about 2 per 

Company 40. — About 5 per cent of the solder type of bond will need replacing at the 
end of first year and about 20 per cent at end of second year, after which the depre- 
ciation is much faster. We do not believe that the average life of installation of solder 
bonds will run much over 3 years. 

The average life of the pin-terminal type of bond is from 4 to 6 years; the depreciation 
of contact, however, is very marked before the elapse of this period. The resistance 
increases about a third at end of first year and has about 50 per cent more resistance 
after the third year than at installation, while at the end of the 4 years, besides the 
increase in resistance, we have found about 40 per cent of bonds out of rail. 

The welded bond shows no increase in contact resistance during our 11 years of use. 
If joints are not kept tight, failures will occur, due to strands breaking. It is hard for 
us to set a percentage of failure, for we have found that on lines where the track is kept 
in good condition the bonds have not depreciated in any way. Up to the present 
time we have replaced about 1 per cent of these bonds a year, due to joint deprecia- 
tion, theft, and careless workmen. 

Company 41. — Practically all of the compressed bonds are giving good service after 
9 years of operation, the life of the road. Have had some trouble with electric-welded 
bonds due to faulty installation during construction. 

38 Technologic Papers of the Bureau of Standards 

COMPANY 42. — Since our company began operation, February, 1908, from the main- 
tenance standpoint we have replaced 7357 bonds. 

This is on an installation of 33 000 twin-terminal bonds and 
amounts to 3 per cent per year. 

(e) Question 7. Intervals for Cross Bonding. — The follow- 
ing figures show the intervals at which the companies cross-bond 
in city streets: 


2 companies 100 to 200 

4 companies 300 

16 companies 500 

4 companies . 600 

10 companies 1000 

1 company 1500 

2 companies 3000 

As a rule these figures apply to both single and double track, 
although four companies bond the two tracks on double-track 
construction at greater intervals than they do the two rails on each 
track. Eight companies cross-bond open and suburban track 
at greater intervals than city track. One company employs no 
cross bonds, owing to the use of block signals. Two companies 
install cross bonds directly beneath trolley feed taps in order 
that they may easily be located at any time. 

(/) Question 8. Size of Cross Bonds. — -With but few excep- 
tions the companies appear to have no other than an arbitrary 
basis for determining the size of cross bonds. Twenty-one of the 
42 answering use 4/0 copper exclusively, while 1 1 proportion their 
cross bonds to accord with the load on the rails and the distance 
from the power house. One company cross bonds with copper 
20 per cent greater than the combined capacity of feeders and 

(g) Questions 9 and 10. Bonding Around Special Work. — 
With but a single exception the 42 companies use supplementary 
copper around their special work in addition to bonding through 
it. The type of bond terminal used for this and cross bonding is 
with nearly all companies the same as that employed on their 
straight work, extra lengths' of copper being soldered to the 
terminals when needed. A few roads using compressed-terminal 
bonds on straight work resort to pin-terminal bonds on special 
work, owing to the difficulty of using a compressor in the sharp 

Rail Joints and Bonds 39 

angles of the frogs. With regard to this feature, Company 3 
states that they now order their frogs with an extra foot of rail, 
thereby permitting the use of the compressor in all cases. 

(h) Question ii. Theft of Bonds. — Thirty-four of the 42 
companies report the loss of bonds by theft in varying extents. 
With 16 of these the trouble has been chronic, while with 18 it 
has been slight or infrequent. It has been much more serious on 
soldered bonds than on other types, owing to the comparative 
ease with which these can be removed. Cross bonds and jumper 
bonds have been stolen in great numbers, even where they were 
buried in the earth. On open construction the only way of safe- 
guarding long bonds appears to be to concrete them in. One 
company reports that after losing many cross bonds this method 
of protecting them was resorted to and so incensed the thieves 
whose revenue had thus been eliminated that they cut and muti- 
lated the bonds at the terminals, apparently for spite. Theft of 
copper is not confined to bonds, but trolley wire and even No. 8 
ground wires have been known to disappear overnight. Com- 
pany 40 reports that in the past 1 1 years 350 welded bonds and 
3500 soldered bonds have been stolen. Company 42 states that 
in 7 years 800 twin- terminal bonds have been stolen, the total 
number installed being 33 000. These are the only roads giving 
definite figures on theft of bonds. 

(1) Question 12. Grade of Labor for Bonding. — The ques- 
tion as stated is ambiguous, owing to the different interpretations 
placed on the words "skilled" and "unskilled." With some 
companies a skilled laborer is one who has had considerable 
experience in applying bonds, although he may not have had 
training as a machinist. With other companies only experienced 
mechanics would be considered as skilled. In view of the exist- 
ence of this ambiguity, the answers received to this question do 
not always give a definite idea of the grade of labor employed. 

The following brief summary of the answers is given with no 
attempt at interpretation: Fifteen companies use skilled laborers; 
10 companies use unskilled laborers; 8 companies use a skilled 
foreman and unskilled laborers; 7 companies use experienced la- 
borers; 2 companies use skilled laborers for welded bonds and 
unskilled laborers for mechanically applied bonds. 

40 Technologic Papers of the Bureau of Standards 

(j) Question 13. Bonding Tools. — As the answers are not 
complete, it is impossible to say how many companies use the 
hydraulic and how many the screw compressor, or to what extent 
power drills are used in preference to hand drills. No summary 
of answers to this question will be attempted. 

(k) Questions 14 and 15. Drilling. — Twenty-one companies 
use no lubricant of any kind in drilling holes for bonds, 4 use water 
only, 8 use soapy water, 2 use lime or soda water, 3 use oil, 19 
roads grind drills by hand, and 13 roads grind drills by machine. 

(I) Question 16. Cost of Bonding. — The cost of bonding is 
extremely variable, depending upon the locality, the price of 
copper, whether holes are punched at factory or drilled on the 
job, whether joint plates are applied by bonding gang, whether 
work is new or repair, whether traffic interferes or not, and upon 
a great many other factors. As average values, therefore, would 
have little significance, no attempt at tabulation of bonding costs 
will be attempted. A few specific quotations giving conditions 
under which the work was accomplished will be of much more 
value and are therefore given in this form. 

Company 2 gives the following unqualified, figures: 

10-inch 4/0 compressed terminals. . . $0. 65 

8,K-inch. 4/0 electric weld 50 

36-inch 4/0 welded to web 95 

8-inch twin terminal 55 

Company 4 gives the cost of — 

Standard leaf bonds, soldered $0. 45 

250 000 cir. mils cross bonds soldered 2. 00 

Company 5. — Cost of installing one 9-inch flexible-crown pin-rail bond we estimate 
as follows: 

Bond $0. 40 

Labor .30 

Tools 05 

Total 75 

Company 6. — Compressed terminal, new work. Holes drilled at factory: 

Labor $0. 04 

Bond 417 

Miscellaneous 0045 

Total 4615 

Rail Joints and Bonds 

Company 7. — Operating full day, installing twin-terminal bonds: 

Bond $0. 35 

Labor 15 

Gasoline and solder 03 

Miscellaneous 0035 


Total 5335 

Company 8. Bonding-Installation Costs 

Rail where bond was 

} Head 




Type of bond 

\to head 




to head 



to web 






Length of bond 
















Size of bond 


Cost of bond 

Labor cost (installing 

Labor cost, removing 
and replacing splice 



















Labor cost, paving 








Paving materials 


Interest and deprecia- 
tion on tools and ap- 

Cost of electrical 










Joint material 


Miscellaneous sup- 
plies and expense. . . 











Total cost 













Company ii. — Cost of installing twin-terminal bonds in quantities n cents to 15 
cents each, exclusive of cost of bond. Cost of installing pin or compressed terminal 
bonds where holes are only reamed, 25 cents per joint where two bonds are installed. 

Company 13. — The cost of installing the welded bond is about 20 cents each, 
exclusive of cost of bond, renewals under traffic about 36 cents, and cross bonds 64 
cents. The latter are placed on the web of the rail. 

42 Technologic Papers of the Bureau of Standards 

Company 14. — Total cost of installing thermo bonding 100 new 36-inch 4/0 bonds 
in open track: 

100 new 4/0, 36-inch compressed bonds $82. 46 

Labor, drilling and applying new bonds, 2 men and 1 foreman, 

3 days 20. 00 

25 pounds thermite, at 30 cents 7. 30 

]/ 2 pound powder, at $1.25 ' .63 

Cost of tools and depreciation 2. 00 

Total cost 112. 59 • 

Cost per bond 1. 13 

100 new 10-inch 4/0 compressed bonds in open track, itemized 

as above (per bond) 76 

(Cost of rebonding with 36-inch, 4/0 compressed bond same 

as for new work, with credit of $0,105 P er bond for scrap.) 

Cost of rebonding exposed track with 10-inch 4/0 compressed 

bond (each) 89 

Scrap credit (each) 07 

Net cost (each) 82 

Cost of rebonding around joints in brick pavement 10-inch 4/0 

compressed bonds (each) 1. 04 

Scrap credit (each) 07 

Net cost (each) 97 

Company 16. — Total cost per joint of double bonding with 4/0 compressed- terminal 
concealed bonds itemized as follows: 

Two 4/0 bonds $0. 80 

Labor drilling holes, applying bonds, and plates on 7 and 9 inch 

rail 50 

Depreciation at 1 5 per cent and interest at 5 per cent on $500 equip- 
ment equals fixed charges per year, $100. 

On basis of 700 joints per year 14 

Electric current, miscellaneous supplies 06 

Total 1. 50 

Same for T-rail construction 1. 42 

Company 17. — The average cost of installing a 9^-inch brazed bond, inclusive of 
paving, is $0. 6392 . The labor costs are from records of 57 10 bonds installed in 19 13 and 
1914. The highest labor cost was $0.85 per bond. This was at a point where four 
bonds only were installed and the traffic was heavy, so it was necessary to do this work 
at night. The lowest cost of labor per bond was $0.10, which was on open track where 
the service could be kept off most of the day. The cost of opening and replacing 
pavement is $0.25. 

Rail Joints and Bonds 43 

The following figures are from Company 20 and are based on 
1000 bonds installed. All types are soldered bonds: 


38-inch cable bond $0. 3567 

Type B. B. bond in paving 59 

Type A, 750 Mem 1. 59 

Horseshoe bond 46 

Type C twin terminal soldered 808 

B. B. bond open-track work, includes removing and replacing 

plates 66 

38-inch Clark 500 M c. m. bond 1. 94 

Company 23. — Our cost for installing 19 818 electric-weld bonds in 1912 wasasfollows: 

Labor per bond $0. 13004 

Material per bond 3138 

Total per bond 44384 

This does not include interest and depreciation of bonding car and tools nor cost 
of electric energy used. 

The cost of labor for the compressed-terminal type bond will average from $0.50 
per bond on new work to $1 . 2 5 per bond on old work. The cost depends largely upon 
the work required in taking up and replacing pavement. 

Company 25. — (1) Single bonding, 70-pound A. S. C. E. rail, 4/0, 10-inch, com- 
pressed-terminal bond, no traffic, bond holes punched, electric current not used, 
joints in large numbers. 

Removing and replacing angle bars $0. 08 

Reaming bond holes 07 

Alloying bond holes 04 

Compressing-bond terminals, painting and adjusting 109 

Total 299 

Superintendence 03 

Tools, 2 per cent 006 

Material : 

One 10-inch bond 363 

Alloy 05 

P and B No. 1 paint 012 

Total 425 

Store expense, 2 per cent 008 

Total cost per joint 768 

(2) Double bonding, same as (1), $1,445 P er bond. 

(3) Single bonding, reconstruction, 141-pound P. S. Co. No. 263 rail, 450 000 cir. 
mil., ribbon compressed-terminal bond, bond holes punched, traffic under 20-minute 
headway, electrical energy not used, joints in small numbers. Itemized as above, 
$1,261 per joint. 

44 Technologic Papers of the Bureau of Standards 

{4) Single bonding, repairs, oiled macadam pavement, 70-pound A. S. C. E. rail, 
4 o, 10 inch, compressed-terminal bond, holes not punched, traffic under 10-minute 
headway, electrical energy not used, track fastenings in good condition, joints in small 
numbers. Itemized as above, $1.38 per joint. 

(5) Single bonding, repairs, asphalt pavement, same as (4), $1,971 per joint. 

(6) Double bonding, same as (4), $2.21 per joint. 

Company 26. — Compressed-terminal, soldered 9-inch bond, $0.55;. labor, $0.35; 

solder and gas, $0.10; total, $1. Twin terminal, $0.42; labor, $0.22; total, $0.64, 

exclusive of paving charges. 

Company 27. — Complete cost of installing oxy-acetylene-welded bonds of 250 000 

cir. mils capacity in large numbers is between 50 and 60 cents per bond, including 

cost of tools and machinery. 
Company 31. — Brazed bonds, $0.56 in quantities; $0.70 to $1.55 for renewals. 

Compressed terminal, $0.69 in quantities; $0.75 to $2.30 for renewals. 
Company 32. — The cost of bonds given herewith includes the cost of the bond 

and the labor of installation for open-track work. A good many of the bonds are 

made in the company's shops, and the copper cost at which they are figured is 14 cents 

per pound. No interest and depreciation on the cost of tools used is taken into account, 

as these figures reduced to a per bond basis are considered negligible. This also 

applies to the cost of power. 

72-inch compression bond $1. 25 

39-inch compression bond 95 

33-inch compression bond 90 

22-inch compression bond 75 

22-inch head-compression bond , 40 

Twin terminal 61 

9 inch crown, concealed 98 

7-inch soldered 65 

Company 33 submits the following table: 

Company 33. Bonding Installation Costs 

Size of bond 

Form of copper 





4/0 (one joint) 

Strand exposed 

. do 








Elevated . . 


Elevated . . 


4/0 (one joint) 

\ $2.30 

300 000 cir. mils 




300 000 cir. mils 


[ 1.99 

425 000 cir. mils 

Strand semiconcealed. 

Ribbon exposed 

Ribbon concealed 

Strand concealed 




550 000 cir. mils 


400 000 cir. mils 


400 000 cir. mils 


300 000 cir. mils 


300 000 cir. mils 

Ribbon concealed 

Solid concealed 


437 675 cir. mils 


Rail Joints and Bonds 45 

Company 35 states that the labor cost of installing a pin- terminal 
bond is 35 cents. This figure is significant when considered in 
connection with the table giving percentage of failures on this 
road. More will be said later regarding the methods employed 
by this company in installing pin-terminal bonds. 

Company 38. — Installing new bonds, large quantities: 

Drilling and reaming f|-inch holes $0. 05 

Compressing bond 05 

One 9-inch bond * 40 

Electric current 06 

Depreciation 05 

Total 61 

Renewing bonds in pavement: 

One 9-inch bond $0. 40 

Six T bolts 24 

20 paving bricks 40 

Cement and sand 10 

Electric current 02 

Labor 1. 68 

Depreciation 28 

Total 3. 04 

Company 40. — 300 000 cir. mils solder bonds cost us 63 cents to install in large num- 
bers and about 40 cents each on renewals. 

The pin-terminal bonds cost us approximately $1 each to install in large numbers 
and about $1.75 each on renewals. The renewal price covers labor in taking off and 
putting back plates and using one-half new bolts. 

The electric-weld bonds cost us 43 cents each to install on renewals. The above 
prices include all labor, current used, supplies, and the interest and depreciation on 
the tools and apparatus. No paving charges are included. 

Company 42. — Twin-terminal bond: Bond, 35 cents; labor and tools, 15 cents; 
total, 50 cents. 

(m) Question 17. Causes of Failure. — The companies are 
almost unanimous in stating that the principal cause of bond fail- 
ures is vibration, resulting from loose rail joints. Two companies 
have given corrosion as the chief cause of failures while five state 
that many bonds fail from external causes, such as traffic and 
injury from workmen. Other causes of failure enumerated are 
expansion and contraction and poor workmanship. 

The following characteristic quotations are selected from answers 
submitted to question 1 7 : 

Company 5. — The only cause of failure of pin-terminal bonds is either defective 
installation at the start or the loosening of bolts and the vibration of the joint. 

46 Technologic Papers of the Bureau of Standards 

Company S. — From our experience the causes for the failure of rail bonds have been 
twofold: (i) Improper maintenance of the mechanical condition of the rail joint, and 
(af) carelessness on the part of the laborer in the installation of the bond in question. 

Our experience has been that 70 per cent of the defective bonds requiring renewal 
will be found at rail joints, whose mechanical condition is very poor. This has given 
the conclusion that a large percentage of bond failures is due to the poor mechanical 
condition of the joint. 

Carelessness in the installation of the bond is brought about in several ways: (1) 
Imperfect drilling of the hole; (2) not wiping the hole clean of the lubricant used for 
the drill; (3) not cleaning the terminal of the bond thoroughly; (4) improper use of the 
bond press or other tools for installing the bonds; (5) crushing the bond between the 
splice bars and rail, when the splice bars are installed, and in the case of the brazed 
bond, " burning the bond" by allowing the welding or brazing temperature to become 
too high and thus making a poor contact between the. bond and rail. 

Company 15. — The causes of failure are usually two. The contact corrodes or the 
rail splits through the bond hole. Frequently a split rail is shown up by the bond 
testing before any sign of split is seen at the head of the rail except probably the joint 
is a little loose. 

Company 20. — We have had only about five failures with the type C twin-terminaf 
soldered bond since we started using them and in all cases failures were due to ballast 
becoming wedged between the bond and the rail and a defective wheel forcing the 
rock down and destroying the bond. 

Company 22. — Many compressed-terminal bonds have become loose in the rail 
through electrolytic action caused by poor contact. Have had no trouble of that kind 
with pin-terminal bonds. Most of our bond failures have been due to broken bonds, 
caused by joint working. 

Company 24. — Brazed bonds are often too short and break when the joints become 
low or slightly loose. Compressed-type concealed bonds: (1) Due to loose joint; (2) 
splice plate binding bond; (3) poor installation, not having bond terminal bright and 
not compressing terminal enough or too much. 

Company 28. — We have found that the principal cause of failure on all bonds other 
than the brazed has been on account of poor contact. With the brazed bonds, how- 
ever, the contact remains perfect and failure only occurs when the bond breaks. 

Company 30. — Soldered bonds crack off from contraction of rail or from weather 
conditions. Stranded compressed-terminal bonds pinch and break at the end of the 

Company 39. — The most prevalent cause of failure of concealed bonds is breaking 
of strands due to excessive vibration. Some other causes are: (1) Terminals broken 
by hammer blow of repair man; (2) strands broken by being squeezed by splice bar; 
(3) strands burned by accidental short-circuit of third rail; (4) strands broken by 
replacing of bolts in splice bars. In exposed bonds failures have been very rare and 
are included in (1) and (3) of above causes. 

Company 40. — (a) We find that the solder bonds fail due to the fact it is hard for even 
an experienced operator to obtain uniform results in installing. After installation, 
the action of the current seems to crystallize and deteriorate the solder, causing the 
terminals to drop off. The mechanical union between the bond and the rail is not of 
sufficient strength so as to prevent bonds being knocked off with a hammer and stolen. 
The strands break, of course, on loose joints. 

Rail Joints and Bonds 47 

(b) Pin-terminal bonds fail due to the corrosion of the terminals and the loosening 
of the same and strands breaking on loose joints. We believe that the corroding of 
the terminals is hastened in a new installation due to the fact that it is almost impos- 
sible to drill holes in the rail of an absolutely uniform size. This is also hastened by 
the unequal coefficient of expansion of the copper and steel during temperature 

(c) Electric-weld bonds fail by strands breaking on loose joints. We have had very 
few failures due to poor workmanship, and have had no terminals depreciate or drop 
off when properly installed. 

Company 41. — It has been our experience during the six years of service on this 
property that, with the use of the compressed-terminal concealed bonds, drilling the 
holes in the web of the rail, making a total of four holes in the end of each rail, weakens 
the rail at this point to such an extent that any undue strain brought to bear on this 
point will cause the rail to break. Have had eight breaks in the six years, and with 
the exception of two have always found the fracture followed down through one of 
the bond holes. 

(n) Question 18. Resistance of Bonds, New and Old. — 
Information on this point is somewhat meager and inconsistent. 
Only a few companies were able to give figures, and these are quoted 
below : 

Company 7. — A new twin-terminal bond will equal about 3 feet of rail, when the 
old ones will average about 4^ or 5 feet. 

Company 8. — The average resistance in terms of rail length for the different types 
of bond joints is rather an indefinite figure, due to the number of variables that enter 
into same. A joint newly bonded with a 4/0 bond will have a resistance equivalent 
to approximately 3 to 6 feet of adjoining rail, depending on the size of the rail. In 
about two years this will increase to about 5 to 8 feet, remaining constant at that 
amount for about six or eight years, when its resistance increases rapidly. 

Company ii. — With reference to twin-terminal and compressed-terminal bonds, 
we have found no difference in resistance between old and new bonding, so long as 
the bonds remain unbroken and their terminal contacts are unimpaired. 

Company 14. — The average resistance of new compressed-terminal bond is about 6 
feet of rail. The average resistance of an old compressed bond is about 13 feet of rail. 
The resistance of brazed bonds after being installed five years is about 5 feet of rail. 

Company 16. — Our experience has been that there is very little change in the 
resistance of a bond if properly installed. As an example will give the results of a 
test made within the last 10 days. 

The piece of track on which the test was made was constructed in 1907. Weight 
of rail, 85 pounds; car service, 10-minute intervals; construction, concrete ballast 
with subballast of crushed rock; average current per rail, 300 to 400 amperes. 

The results of the test show the drop in millivolts on 24 joints, including 3 feet of 
rail, as compared to the drop on the 3 feet of adjacent rail. The figures range from 10 
to 15 for the solid rail and from 10 to 18 for the joint. The average resistance of the 24 
joints is about 10 per cent greater than that of 3 feet of solid rail. 

The joints are bonded with two 400 000 cir. mils bonds per joint. The copper 
equivalent of these rails is about 850 000 cir. mils. We do not have on record the 

48 Technologic Papers of the Bureau of Standards 

readings on the joints when new, but considering the amount of copper in the joint 
and the copper equivalent of the rail we do not believe the resistance of the bonds 
could have changed but very little during seven years of service. 

Our experience in compressed-terminal bonding indicates that the essentials for 
good results are to have the holes drilled of uniform diameter, free from all traces of 
any lubricant, bond terminals machine finished, in order to assure a driving fit; 
finally, the maximum amount of compression possible with tools available for this class 
of work. 

Company 19. — We consider that two 4/0 compressed- terminal bonds with a contin- 
uous rail joint has the equivalent resistance of 4 feet of 70-pound rail when new. Age 
does not affect this if the track joint is kept in good condition. 

Company 26. — Tests of compressed-terminal bonds and twin-terminal bonds are 
practically the same, 4 to 4% new, 5 to 6 when old. 

Company 34. — Two 36-inch 4/0 bonds, pin-terminal, on 60-pound T rail, when 
new resistance equals 8 feet of rail. When in six years, resistance equals 10 to 12 
feet of rail. Compressed terminals show about the same resistance. 

Company 35. — Average resistance of pin-terminal bonds on track rail equals 18 
to 20 inches new, 36 to 40 inches old. 

Company 37. — In the use of brazed bonds there is a remarkable similarity between 
the resistance of the bond when first applied and when old, which indicates the great 
advantage of this type of bond over all others. 

Company 40. — (a) The 9-inch 300 000 c. m. soldered bond when first installed on 
70-pound rail equals 8 feet of rail. We lose about 5 per cent of the bonds during the 
first year, and about 20 per cent the second year. 

(b) The 10-inch 4/0 pin terminal when first installed on 85-pound tee rail equals 
1% feet of rail. The same bonds are equal to 9^ feet of rail at the end of first year, 
about 10X feet of rail at end of third year, while at the end of four years we have 
found 40 per cent of the bonds out of the rail. 

(c) The short 4/0 electric-weld bonds are equal to 4.9 feet of rail when first installed 
and have shown no terminal depreciation during our 1 1 years of use. 


Of the 42 companies enumerated heretofore but 17 employ- 
other than bolted joints, and the data submitted by these roads 
apply for the most part to old types of welds which are now either 
practically obsolete or which have recently been so modified and 
improved that records of failures during the past years are unre- 
liable and misleading as an indication of the performance of modern 
welded joints. Unfortunately, but few of the larger traction 
companies whose experience would be most valuable on this sub- 
ject are included in the list of those answering our circular, and 
while many of these have been visited and interviewed regarding 
their experience with rail joints, very little in the way of statistics 
from them is at hand for tabulation. 

Rail Joints and Bonds 


It is obviously unwise to generalize or base conclusions and 
percentages on data compiled from so few as 17 reports, represent- 
ing as they do but a small percentage of the total mileage equipped 
with welded joints. The tabulations and quotations which follow, 
therefore, should be considered as representing special rather 
than average conditions. They will, however, be used in a later 
part of the report in connection with material obtained from other 
sources as the basis for some general conclusions and recommen- 

(a) Question i. Number of joints in use. — Table 6 gives the 
total number of welded as well as bolted joints reported by the 
several companies answering, as well as the number of companies 
reporting the use of each type of joint. 

As with the bonds, only a fraction of the total number of joints 
represented have been reported, owing apparently to the lack of 
records. The questions on welded and other types of joints were 
intended to cover only electrically continuous joints. Nine 
companies, however, gave figures on bolted joints as well, which 
accounts for the comparatively small number reported. 

Number of Rail Joints Reported 

in use 

nies re- 

in use 

nies re- 

Electrically continuous joints : 
Cast welds 

149 716 
46 849 

50 000 

8 490 


Bolted joints— Continued. 


Thermite welds 

20 000 


Electric welds 

Bonanzo and Duquesne. . . 

Arc welds 


Clark joint 

2 000 

159 402 

32 671 

9 336 


Nichols joint 


268 052 


Bolted joints: 

68 669 
7 700 


301 925 

Continuous joints 

Weber joints 

150211°— 19- 

v so Technologic Papers of the Bureau of Standards 

(/)) Question 2. Types of Construction. — With but few ex- 
ceptions welded joints have been installed only in city paved 
streets with concrete or stone ballast. A few companies report 
the use of welded joints in macadam or earth streets and one on 
open track. 

(c) Question 3. Electrical Efficiency of Joints.— Ten com- 
panies reported the efficiencies of their welded joints to be 100 per 
cent or better. One company gave the efficiency of cast welds to 
be 67 per cent. Six companies had no data on this question. The 
electrically-welded joint was reported by various companies from 
100 per cent to 170 per cent efficient, electrically. 

(d) Questions 4 and 6. I^ife of Joints and Causes of 
Failure. — The following quotations, taken from answers to ques- 
tions 4 and 6, are given in preference to a summary which would 
not be satisfactory : 

Company 3 . — Our experience shows that the life of our rail on good track is from 
15 to 20 years under our ordinary traffic conditions, say one 20-ton car every 3 or 4 
minutes per 18 hours. On this type of track 95 per cent of our 5000 cast-welded joints 
have lasted the entire life of the rail. The electrically welded joints have been in 
only two years. We shall be disappointed if 95 per cent do not last the life of the rail. 

Where the continuous rail joint has been applied we had no renewals for the first 
two years. From the third to the fifth year about 20 per cent of the joints required 
renewal. At the end of 10 years the renewal was total; that is, all of the joints had 
been renewed. With ordinary splices under the same traffic conditions as those 
under which the continuous joints were used, all of the splices had to be renewed 
within five years. 

The above is an opinion in the absence of absolute data, but is in a general way 

The usual cause of failure in every joint is the deflection of the joint due to the 
load, causing an ultimate pounding of the joint. The final result is a cupped place 
on the receiving rail and looseness due to wear, the wear being on the fishing surface 
of both the rail and the splice. This is true for every type of joint. 

When rail is first welded either electrically or by means of cast weld about 1 per 
cent of breaks will occur within the first 60 days. We expect such breaks and consider 
them necessary to permit the rail to adjust itself to conditions, and the repair of these 
breaks should be included in the cost of the first installation of the welded joint. 

Company 4. — The average number of cast- weld joints replaced in repairs each 
year is about 1000, the total number of joints in service being about 42 000. A great 
number of joints were taken out when tracks were replaced in 1913 and 1914 where 
the joints were originally installed in about 1895, thereby indicating that the majority 
of the joints will last as long as the rail, the failure being due to the rail itself and not 
to the joints. 

The most prevalent cause of failure in the cast- weld joint is the loosening up of the 
rail in the joint. 

Rail Joints and Bonds 51 

Company ii. — Riveted and thermite- welded joints (Clark joints) have been in 
use but three years. So far we have had no failures and no defective joints where 
applied to new rail. Every indication points to a life of this type of joint equal to the 
life of the rail. 

Lorain electrically welded joints have shown but from 1 to 5 per cent, or more 
failures per annum, depending upon the condition of the rails. 

Bolted joints will in no case last as long as the rail. 

Our experience with cast- welded joints has been more satisfactory than with bolted 
joints. The percentage of failures is not as great as in electrically welded joints. 

Electrically welded and cast-welded joints have nearly all been applied to partly 
worn track; therefore our experience is probably not a fair guide to similar joints 
applied to new track. 

In electrically welded joints the failures are due to many causes, principally the 
breaking of the rail at the ends of the welding bars, splitting of the web above the 
bars and the depression of the head due to nonsupport. 

Cast- welded joint failures are due to depressed or cupped heads and split webs. 
There are very few cases where joints show any true welding. 

Company 16. — In the 100 thermite joints there have been four breaks in four years 
of service. The rest show serious mechanical wear which will soon require extensive 
repairs or replacement. The life of other types of joints varies with service and the 
care taken of the track. Under our average conditions a continuous type of joint 
should last from 12 to 15 years. We have no definite data in this respect, as this type 
of joint has not been in use long enough to require replacing. 

The thermite- welded joints have failed by breaking due to contraction of rail. 
They also show very rapid mechanical wear, due to the soft material in the weld. 
This latter cause of failure has largely been removed by recent improvements. 

Company 17. — The most prevalent cause of failure of a mechanical bolted joint in 
our case has been due to not drawing the rails together as tightly as possible before 
finally bolting on the joint and the irregularities in the rolling of the rails. 

The few welded joints we had have not given satisfaction, to some extent on account 
of being used on old track. We have never had any on new steel. 

Company 18. — In regard to cast- welded joints would say that they have given 
good service for 15 years and taken as a whole I consider that they have been a success. 
Of course there have been some failures and in the 15 years a great many mechanical 
joints have been substituted for the welded joints, but I judge that this has not 
amounted to more than 10 per cent, and would possibly be as low as 5 per cent of the 
total number. After 13 to 15 years service these joints are still failing at the rate of 
probably 1 per cent or 2 per cent per year, and of course the failures in previous 
years have been less than that amount. 

In regard to mechanical joints I believe that a very much larger per cent of these 
have failed or required repairs. What the percentage is no man can tell. A large 
percentage of those failures I attribute to the fact that a great many of the mechanical 
joints were not put on with care that should have been exercised, nor with proper 
material used in the bolts. Recent experience would indicate that mechanical 
joints when properly applied could be made to hold with as small percentage of 
failures as can be obtained with welded joints. 

The joints made with the thermite welds have been used only on compromise joints. 
The percentage of failures of these has been negligible. 


Technologic Papers of the Bureau of Standards 

No attempt has been made to replace welded joints which have failed. In cases of 
that kind we have always substituted mechanical joints. 

Company 19. — The life of a joint nowadays will in most cases equal that of the rail. 
And while the joint may "pound" within five or six years, depending on the nature 
of the traffic, they can be built up and smoothed off with electric welders and grinders 
so that the joint can be maintained as long as the rail itself. 

The most prevalent cause of failure is the giving away of the track foundation owing 
to water leaks saturating the soil under the track foundation. 

Company 22. — Failure on 1600 thermite welds, late type, not over 1 per cent per 
annum, no exact record, however. 

On 800 arc welds no failures to date where joints were properly welded. 

Company 27. — The failures of cast- welded joints per year is less than three-fourths 
of 1 per cent of the total number of joints in use. 

The usual cause of failure on cast-welded joints results from either imperfect weld 
or from excessive pounding at the joint due to imperfect surfacing or wear. 

Company 32. — Cast welds: The castings crack, the rails pull out of the castings, or 
the receiving rail wears out due to the softening of the rail by the welding heat. The 
latter is the most prevalent cause of failures. 

Channel-bar joints and angle-bar joints: Loose bolts are probably the cause of the 
greatest number of failures in these joints. If these joints are not properly fitted to the 
rail or bolts are not kept tight, the plate is bent or deformed under the rail head on 
the receiving side of the joint. 

Nichol's composite joint: No failures have occurred since these joints were installed 
three years ago. 

Company 34. — The 8 miles of electrically welded joints were installed in 1906 and 
are in good condition at present. Only seven joints have broken. 

This is on suburban road with expansion joints every 1000 feet. 

Company 37. — Have had but three failures on 50 000 Clark joints, standard since 

(e) Question 5. Cost of Joints.— The following cost data are, 
in some instances, of little significance, as they are given without 
information as to weight of rail or as to whether the cost includes 
the bonding of the mechanical joint or the renewal of pavement. 
They are tabulated under the various types of joints for which 
costs were given and include both labor and material. 














62-pound T rail 





56-pound, 85-pound, and 91-pound rail 


Rail Joints and Bonds 


CAST WELD— Continued 





70-pound rail 










32 . 


32 . 




9-pound to 125-pound girder rail 

80-pound girder rail 

70-pound, 7-inch girder rail 

70-pound rail . 

Old type on 60-pound to 116-pound girder rail. 

Average cost 









60-pound to 106-pound rail. 
116-pound girder rail 

.contract price.. 



Average cost. 

9-inch rail 

7-inch rail 

100-pound T rail. 







Using nld angle bars - 






$4. 60-$5. 00 



54 Technologic Papers of the Bureau of Standards 















7-inch and 9-inch 

Paved streets 

Open track k 

12-hole plate, including bond 

7-inch T rail 

Including two 4/0 bonds 

40-pound, 56-pound, and 60-pound T 

70-pound T 

72-pound T 

80-pound high T 

116-pound, 7-inch girder rail 


80-pound T rail 

100% JOINT 

70-pound T rail 


70-pound rail 

7-inch rail 

9-inch rail 

4-hole plates on 91-pound, 7-inch girder rail. . . 

35-pound T rail 

40-pound T rail 

56-pound T rail 

60-pound T rail 

72-pound low T 

80-pound low T 

9-inch rail 

7-inch rail 

100-pound T rail 

80-pound T rail 

60-pound T rail 

40-pound T rail 





(/) Question 7. Temperature Variation in Weuded 
Joints. — The tabulated answers to this question as given in the 
following statement are no doubt largely based upon guesses or 
estimates. In no case is it stated that they are taken from expe- 
rimental observations. 

Rail Joints and Bonds 55 

Company 3. — All welded joints are in paved streets where the rail and track is 
gripped and firmly held by the pavement. The temperatvre of a rail buried in most 
pavement in the summer time rarely exceeds 6o° average temperature in our climate. 
Also the rail buried as above seldom goes below 30 , no matter how cold the atmos- 
pheric temperature is. The maximum temperature variation of a buried rail in our 
climate is probably not greater than 40 . 

Company 4. — Eighty degrees F to 15 F. 

Company 10. — Ninety degrees F to io° F. 

Company 11.— Ninety degrees F to 25 F. The rail never gets as cold or as hot as 
the air temperature. 

Company 16. — The maximum change in temperature of atmosphere is about ioo°. 
The welded joints, being embedded in concrete, are subjected to a considerably less 
range, probably not over 50 . 

Company 24. — Air temperature o° to ioo° F; ground temperature 30 to 70 plus or 

Company 25. — Sixty degrees F maximum range. 

Company 26. — Ninety degrees F to 20 F. 

Company 28. — Fifty-four degrees maximum range. 

Company 30. — Seventy-five degrees F maximum range. 

(g) Question 8. Expansion Joints. — Eight of the 17 com- 
panies use no expansion joints whatever. Those that do are 
quoted below. 

Company 17. — Special expansion joints of a modified slotted angle bar type are used 
on bridges and viaducts and we also use slotted angle bars in open track at intervals. 

Company 22. — On all lines we leave a joint about every 500 feet, which joint is con- 
nected by the usual angle bars and bolts. Our experience is, however, that these 
joints do not move any more than the welded ones, the friction of the pavement hold- 
ing the rail rigid. On one line 7^ blocks long we installed expansion joints with 
slotted bolt holes. This line is partly planked and partly dirt filled to the top of the 
rail, and though most of the joints have broken none of the expansion joints have 

Company 2^. — With welded joints we use continuous plates every 500 feet and at 
special work. 

Company 25. — Approximately every 400 feet continuous bonded joint. We are 
not welding any open track. 

Company 34. — Expansion joints in the electrically welded track are placed 1000 
feet apart. In other track none, except on open track, where }i inch to yi inch is 
allowed, depending on temperature when laid. 

Company 37 . — Do not use expansion joints on regular work. Only expansion joints 
we have are on long viaducts, where rail is fastened directly to bridge members. 


That the problem of rail-bond maintenance is largely that of 
joint maintenance and becomes serious only where loose rail joints 
exist is attested by a large majority of the operating companies 

56 Technologic Papers of the Bureau of Standards 

enumerated in Section II of this paper as well as by the verbal 
testimony of the engineers of a number of other companies. It is 
apparent, therefore, that the first step in an attempt to reduce 
the present high percentage of bond failures should be toward bet- 
tering the condition of the bolted joint. 

The principal causes of failure of the ordinary bolted joint have 
been frequently discussed and are more or less familiar to the ma- 
jority of electric railway engineers, and will here be reviewed only 

(a) Defective Roadbed. — This includes poor ballast, poor 
drainage, water leaks, and decayed ties. While these defects are 
external to the joint proper, they nevertheless exist and will tend 
to the ultimate failure of the best of bolted joints. 

(b) Nonuniforms y of Rail Sections. — No two rail sections 
will be found to be absolutely identical. The difference will vary 
from an inappreciable minimum to the maximum allowed in the 
specifications adopted by the American Society for Testing Mate- 
rials. These inequalities are the result of wear of rolls and different 
degrees of shrinkage with cooling. When two unequal rail sections 
are joined together by uniform splice bars it is evident, therefore, 
that the bars do not fit with equal closeness the fishing surfaces 
of the two rails at the joint. With the passage of the wheel over 
such a joint one rail will be depressed below the other and pounding 
of the joint will result, particularly if the receiving rail is the lower 
of the two. Continual pounding of a joint eventually develops a 
cup in the receiving rail and rapid depreciation follows if the joint 
is not given proper attention 

(c) Defective Rail Ends. — Rails rolled some years ago by 
some of the steel companies had a distinct dip of the head at both 
ends. When joined together a depression or cup was left at each 
joint which, if allowed to remain, resulted in further wear and early 
failure of the joint. An order of rails received in Worcester, Mass., 
at one time was so defective in this respect that both ends of all 
rails had to be sawed off before they were installed. While this 
condition has not been met with in recent years it has no doubt 
been the cause of a number of joint failures in the past. 

(d) Failure to Grind Joints. — Slight inequalities exist in rail 
heads as well as in the fishing surfaces, so that on newly bolted 

Rail Joints and Bonds 57 

joints a difference in elevation of the abutting rails often exists 
and unless filed or ground down to a perfect surface alignment will 
soon develop pounding and cupping. A number of companies 
now make a practice of running over all newly bolted joints with 
a track grinder and find that the slight expense is well justified 
by the increased smoothness and resulting longer life of the joints. 

(e) Loose Bolts. — This is by far the most universal and preva- 
lent cause of complaint in connection with joint failures and has 
been the source of constant annoyance and expense to practically 
every electric railway in the country. Bolts become loose through 
various causes. They are stretched beyond their elastic limit 
when being tightened and are thereafter unable to accommodate 
themselves to expansion and contraction incident to temperature 
changes and to the vibrations from traffic. As soon as the ten- 
sion on the bolts is thus relieved the joint begins to work, which 
allows the plates to wear and further augments the pounding 
and cupping. It is practically impossible to permanently install 
joint plates before traffic is put on the track. Slight inequalities 
in plates and rails permit their intimate contact only at points, 
and until these high places are worn off or flattened by several 
days' traffic a proper seating of the plates is difficult. A number 
of operating companies now follow up all bolted joint installations 
and take up the slack in the bolts which has developed with a 
few days' traffic. This practice is obviously commendable and 
should be rigidly adhered to under all circumstances. The tight- 
ening of bolts on all exposed joints once each year or at other 
regular intervals is practiced by some companies with a corre- 
sponding increase in the average life of the joints. 

(/) Improved Bolts. — Perhaps the most important improve- 
ment in the bolted rail joint in recent years and the one which 
will do more than any other thing to lengthen the life of the joint 
is the substitution of improved bolts for the more inferior grades. 
Within the last few years bolts having a high elastic limit as well 
as a great ultimate strength have been put on the market by sev- 
eral manufacturing companies and are now coming into general 
use both for special work and straight track. 

A large number of companies was consulted regarding the 
properties and effectiveness of these bolts, and, while a number of 

sS Technologic Papers of the Bureau of Standards 

them stated that trial orders had been placed for a limited number 
or that they were experimenting with them and believed them far su- 
perior to the old bolts, practically no one was in a position to give 
positive testimony based on observations covering any great length 
of time. One notable exception to this was found with the Harris- 
burg Railway Co., which was using improved bolts on a steam 
railroad crossing where endless trouble had been experienced with 
ordinary bolts, including various patent washers and lock nuts. 
After the installation of the improved bolts no trouble was ex- 
perienced and though no lock nuts or other precautions have been 
employed to keep the nuts tight they have shown no tendency to 
work loose since their installation. 

In answer to question E-3-87, in the June publication of the 
A. B. R. A. for the, year 191 5, asking for information as to the 
economy of using special heat-treated bolts in the joints of rail- 
road crossings and special work the following interesting replies 
are quoted: 

A. E. Harvey, Chief Engineer Metropolitan Street Railway Co., Kansas 
City, Mo. — After a careful study of this question for several years the writer became 
convinced that one of the principal factors in the failure of mechanically applied 
joints was the poor quality of material used in bolts which failed in time, not through 
the stripping of threads or breaking, but through a gradual stretching and loosening. 
Over a year ago the use of the high-grade steel bolt with an elastic limit of 75 000 
pounds was begun, and they have given excellent satisfaction. The high elastic limit 
permits of the stretching of the bolt, to some extent, without its receiving a permanent 
set, thus tending to keep the nut tight. Bolts for this work need not necessarily 
be material of alloyed steel or heat treated. There are a number of concerns that 
now manufacture bolts of high-grade steel with an oil finish that answer every purpose 
in track work and that at a very slight cost above that of the ordinary track bolt. 
The above applies not only to the use of bolts in special work and railroad crossings 
but in all track work. 

A. V. Brown, Engineer Maintenance op Way Lake Shore Electric Rail- 
way Co., Sandusky, Ohio. — Have had excellent success with heat-treated bolts at 
crossing frogs and are now specifying them on all crossings. 

H. A. Clarke, General Manager Ithaca Traction Corporation, Ithaca, 
N. Y. — I would advise that I have not made use of special heat-treated bolts for such 
purposes but have used special steel-alloy bolts, having a very high tensile strength, 
and we find that it is economical. The breaking of such bolts is diminished at least 
50 per cent, and in addition the crossings and special work are held more rigidly, 
decreasing the wear on the joints. We have also found it economical to use special 
alloy-steel bolts of high tensile strength for ordinary joints. We have found a great 
many of the ordinary track joints becoming loose, due to the stretch of ordinary bolts. 
This is almost entirely overcome by the use of special alloy bolts. 

Rail Joints and Bonds 59 

J. B. Tinnon, Engineer Maintenance op Way Chicago & Jouet Electric 
Railway Co., Jouet, III. — The first cost of heat-treated bolts is about twice that of 
ordinary bolts, but the fact that they no not stretch as easily will alone save their 
extra cost, because they will not require tightening as frequently. Heat-treated bolts 
are also much tougher than ordinary bolts and are therefore not so easily broken by 
sudden shocks; in fact, they have to be cut about two-thirds through before they 
will finally break off. Since the stretching of the bolt is the cause of most loose and 
broken bolts, the advantage of a tougher bolt that will not readily stretch is very 

George H. Pegram, Chief Engineer Interborough Rapid Transit Co., New 
York, N. Y. — Our experience does not justify me in specifying special heat-treated 
bolts. Our specifications, however, require a bolt of high tensile strength with a 
severe bending test, which requires a high quality of steel. 

This testimony in favor of bolts having a high elastic limit as 
against iron or even carbon-steel bolts is apparently conclusive, 
and the slight additional expense is completely justified by the 
decreased maintenance and additional life of the bonds and joints, 
to say nothing of the improved riding properties of the roadbed. 

(g) Joint Plates. — The character of joint plates affects the life 
of rail bonds in so far as they have to do with the general condition 
of the joint with respect to deflection and vibrations and also as 
they tend to restrict the natural vibrations and movements of 
concealed bonds. 

(h) Room for Concealed Bonds. — A concealed bond, i. e., 
one installed under the joint plate, is ordinarily subjected to two 
kinds of vibrations. It is bent or deflected by the deflection of 
the joint under the car wheel, and it is lengthened and shortened 
by the diurnal expansion and contraction of the rails on exposed 
and other joints where such expansion and contraction takes 
place. Concealed bonds are designed to accommodate themselves 
to these motions and do so admirably when not restricted in their 
natural movements by the action of the plates or bolts. 

The vibrations and bending of bonds incident to the vertical 
motion of the joint and rail end is supposed to be taken care of by 
making the bond long enough to withstand such vibrations, while 
the longitudinal motion resulting from expansion and contraction 
is taken up by the crimp which is put in all concealed bonds. 
The ability of a bond to withstand vibrations depends largely 
upon its length. This relation has been made the subject of ex- 
perimental investigation by the American Steel & Wire Co. and 
other manufacturers of bonds. A vibration testing machine 

60 Technologic Papers of the Bureau of Standards 

designed to approximate the motions of a very loose rail joint 
grips the bond terminals and gives them alternate vertical dis- 
placements of any desired amount. With every 125 vertical 
oscillations the bond is lengthened and shortened once through any 
required distance. The following figures are given as results of 
tests on bonds of various length: 7-inch bond began breaking at 
41 000 vibrations; 8-inch bond began breaking at 215 000 vibra- 
tions; 10-inch bond began breaking at 1 279 000 vibrations; 14- 
inch bond began breaking at 7 887 000 vibrations. 

This test gives no indication of the life of a bond on a good joint 
but represents extremely poor track conditions. Converting 
these figures into years of life on a track carrying double-truck 
cars under a six-minute headway we find that the 10-inch bond 
will not begin to fail for five years and the 14-inch bond will remain 
intact for approximately 30 years. That such performances are 
never realized in practice on bad joints and seldom on perfect ones 
requires no argument. 

Without a doubt the chief difference between the laboratory life 
test and performance under service conditions lies in the fact that 
in the latter case the full length of the bond is seldom if ever 
utilized, but is restricted by the squeezing action of the plates, or 
the crowding action of the bolts, and in double bonding sometimes 
by the protruding button of the bond on the opposite side. 

The ordinary splice bar, such as is used on steam roads and 
which was the only type of joint plate available in the early days 
of electric railways, makes no provision for the accommodation 
of the concealed types of bonds and no end of difficulty has been 
experienced by all operating companies on this score, particularly 
with the smaller rail sections. There is general complaint that 
the steel companies have been slow in rolling plates specially 
designed to meet this problem and numerous companies, in their 
earlier installations, resorting to the only available material, 
applied concealed bonds under plates which did not permit of their 
free movement and thereby led to their early failure. 

Fig. 13 shows how the bolts and Fig. 14 how the button of the 
bond on the opposite side might easily prevent any movement of 
the bond between these points and the bond terminal, particularly 
under plates which have not been designed to take care of this 

Bureau of Standards Technologic Paper No. 62 

FiG, 13. — Showing interference of bolts with movement of bond 

FiG. 14. — Showing interference of bond terminal in double bonding 

Fig. 17. — Clip to prevent separation of ribbons 

Fig. 22. — Completed Clark joint 

Rail Joints and Bonds 6 i 

On badly worn and loose joints of the type shown in Fig. 13 the 
entire vertical motion of the bond is confined to the short region 
between the bolts nearest the rail ends and in bonds which are 
hugged tightly by the plates to a still more restricted length. 
Such a condition results in the breaking of the strands, not at the 
bond terminals, as would happen with no restrictions, but at the 
juncture of the rails or near thereto. Even upon well-maintained 
joints in which there is practically no vertical motion the continual 
lengthening and shortening of the bond resulting from expansion 
and contraction, and which is confined to a comparatively short 
length, will produce the same effect. Such failures have been 
more prevalent on open track, where rails experience the full effect 
of temperature variations, than in city streets. This may be par- 
tially the result of better ballast and heavier rails in the latter 
type of construction, but it is quite reasonable to suppose that 
the linear expansion and contraction which takes place on the 
open track is largely responsible for the crystallization and break- 
ing of the strands and ribbons. 

(i) Examples of Bond Failures. — Companies 2 and 14 specifi- 
cally state that expansion and contraction in rail joints is respon- 
sible for bond failures, and Company 16 says that the failure of 
concealed bonds is confined to those joints in which expansion and 
contraction of the rails takes place in the joint. The Boston & 
Worcester Street Railway Co. presents a striking example of this 
type of bond failure on their suburban line between Boston and 
Worcester. It is reported that only a small per cent of the origi- 
nal 12-inch concealed- wire bonds are now in service, a large ma- 
jority of them having failed by the wires breaking in the middle 
near the juncture of the two rails. 

Engineers of the American Railways Co., which operates a num- 
ber of properties, state that they have had difficulty in obtaining 
room for concealed bonds on rails of 60 pounds per yard and 
smaller. On large rails they report ample room and state that 
they will permit of very loose joints without breaking bonds, while 
the slightest motion in the joints of the smaller rail sections will 
quickly result in broken strands or ribbons. 

Some of the bond manufacturers have attempted to meet this 
problem by providing the operating companies with stranded 


Technologic Papers of the Bureau of Standards 

Fig. 15. — Standard joint 
plates, showing inade- 
quate space for bond 

bonds having a triangular section. A sectional view of such a bond 
installed is shown in Fig. 15. While this may be an improvement 
over a bond having a circular section, its very necessity is an ac- 
knowledgment of a condition which requires a more drastic remedy. 
(j) Special Plates. — Some of the larger and more progressive 
companies have called on the steel manufacturers to roll special 
plates for them which have been designed to 
give room for concealed bonds. This has been 
done in some cases and in a few instances such 
plates have become standard with the manu- 
facturers. This is particularly true with re- 
spect to the manufacturers of some of the 
patent joint plates, such as the continuous 
joint and the Bonzano joint. Fig, 16 shows 
a sectional view of continuous joint plates 
applied to a 40-pound rail. It is seen that ample room is provided 
for concealed bonds. The j oint plates recently adopted as standard 
by the Amerian Electric Railway Association were designed to 
give ample clearance for bonds and are now being rolled. These 
standards, however, 
were adopted for 
only the 7 -inch and 
9-inch rails, on which 
the problem of bond- 
ing was not so diffi- 
cult as on the smaller 

(k) Improved 
Joint Plates. — 
Not only do some of 
the improved joint 

1 , f ' 11 * Fig. 16. — Special plates, showing ample provision for bond 

crease the life of concealed bonds by giving ample clearance for 
them, but their ability to better support and maintain the joint 
than the old types of plates is sufficient justification for their use. 
Among the improved bolted joints the continuous joint seems 
to be the most popular with the operating companies. Figures 
submitted in Table 6 show that 19 companies reported the use of 
this type of joint, whereas the largest number using any other 

Rail Joints and Bonds 63 

type of improved plates is 3. The total number of continuous 
joints reported is 68 669, which is more than twice the number of 
all other types of improved joints reported. 

This type of joint plate when properly installed with bolts hav- 
ing a high elastic limit grips the rail so firmly that expansion and 
contraction within the joint is largely eliminated, particularly on 
city tracks. We quote Company 3 with reference to this point: 

We think if the bolts in continuous joints are drawn tight there is very seldom any- 
slipping of the joint, due to expansion or contraction, as this joint grips the rail very 
firmly, so that a well-bolted continuous joint gives nearly the same effect as welding. 

The type of bolted joint referred to under " Mechanical joints" 
in Section II of this paper, wherein a shop fit is obtained by ream- 
ing holes and using machine bolts and in which no expansion and 
contraction is allowed, has, to our knowledge, not been used on 
open track, although welded joints have been used in a number of 
installations for this purpose, expansion and contraction being 
taken care of by expansion joints at regular intervals of about 500 
to 1000 feet. If such improved bolted joints were used on open 
track in connection with expansion joints, a great reduction in the 
maintenance cost of both bond and joint would be affected, to say 
nothing of the economy of operation and improved operating con- 
ditions generally. The ultimate economy of such construction 
would have to be carefully considered for any given project, but 
that it would be fully justified on heavy traction lines is firmly 
believed. If installed according to the best modern practice, bond 
failures would be reduced to a minimum, being relieved of the con- 
tinued lengthening and shortening so prevalent in the ordinary 
joint. Maintenance would consist of occasionally going over the 
joints and tightening the bolts. If in time the rail ends began to 
cup, they could be inexpensively built up by applying new metal 
with the arc welder or acetylene flame and then ground to a true 
surface alignment. Such joints with the comparatively slight 
maintenance here mentioned would undoubtedly have a useful 
life equal to that of the rail and at the same time provide a continu- 
ous and permanent return circuit for the electric current. We 
believe the type of construction here described to be not only 
practicable but of ultimate economy, and urge its adoption by the 
operating companies at least on an experimental basis. 

64 Technologic Papers of the Bureau of Standards 


((7) Comparison of Compressed and Pin Terminal Bonds. — 
Stud-terminal bonds according to our definition on page 12 in- 
clude both compressed-terminal and pin- terminal bonds, and each 
of these types comprise ribbon and wire bonds for both concealed 
and exposed application. 

The tabulation on page 29 shows that 929 600 compressed- 
terminal bonds were reported as against 362 800 of the pin- ter- 
minal type. These figures, together with a majority of testimony 
as to preference, indicate that the compressed terminal is easily 
the favorite among a majority of the operating companies. On 
the other hand, however, must be considered the fact that the 
pin-terminal bond has been adopted as standard and is being em- 
ployed with phenomenal success by a number of the largest oper- 
ating companies, including the New York Central & Hudson River 
Railroad Co. and the Pennsylvania Railroad. While it is not pos- 
sible to say in general that one of these types is better or worse 
than the other it is hoped that a careful analysis of all information 
available will aid in reconciling the differences in opinions regard- 
ing these two types, and establish the fact that each type has 
characteristics and properties which makes it peculiarly adaptable 
for certain classes of work or under certain special conditions. 

One of the arguments put forward in favor of the compressed- 
terminal bond is as follows: The contact resistance between cop- 
per and steel decreases as the pressure increases up to about 30 000 
to 40 000 pounds per square inch. As copper reaches its elastic 
limit and begins to flow at about 20 000 pounds per square inch, 
the minimum contact resistance is not reached with the pin- 
terminal bond since the copper is not confined during the driving 
of the pin, but is free to flow out around the pin, forming a button 
on the opposite side of the rail as is illustrated in Fig. 3. With 
the compressed-terminal bond, it is argued, the copper is confined 
between the terminals of the compressor, and not being able to 
escape is subjected to a pressure limited only by the design of the 
compressor or the diligence of the workmen. 

This argument, which at first may appear to be tenable, is un- 
doubtedly fallacious. It is true that very soft and thoroughly 

Rail Joints and Bonds 65 

annealed copper has an elastic limit of about 20 000 pounds per 
square inch, but upon undergoing a very small amount of manip- 
ulation it rises rapidly to from two to three times this value. As 
the action of the compressor or even the driving of the expanding 
mandrel produces a distortion in the copper more than sufficient 
to bring about this change in the elastic limit it is obvious that 
the pressure required for minimum contact resistance is reached 
in both the compressed and pin terminal type of bonds. 

If there is any difference in the contact resistance of these two 
types it appears to be so slight as to have practically no effect 
upon the total resistance of a bonded rail joint. The average of 
32 tests on each type conducted by the Chicago Board of Super- 
vising Engineers in 191 1 shows the pin-terminal bond to have a 
conductivity of 96.65 per cent of that of the hydraulic-compressed 
bond and 98 per cent of that of the hand-compressed bond. As 
the double contact resistance of a 4/0, 10-inch, copper bond is 
only about 20 or 25 per cent of the total resistance of the bonded 
joint the slight difference in the contact resistance of the two 
types would affect the total resistance of the joint in the order 
of a fraction of 1 per cent, which is so small as to be entirely 
negligible for practical purposes. 

The contact resistance of a stud-terminal bond when newly and 
properly installed is often quite a different thing from the resist- 
ance of the average bond after being subjected to several months 
or years of service. That the two may be quite different is evi- 
denced by the greater part of the testimony recorded in answer 
to question 18, although several companies believe that bonds 
show little if any increase in resistance if the joints are properly 
maintained. The increase in the resistance of a joint in some 
instances may be solely the result of the loosening of the joint 
plates. Tests recently conducted by the Bureau of Standards, 
the results of which are given on page 119 of this paper, show 
that newly bonded and bolted joints have a much lower resist- 
ance with the plates on than with the plates removed, indicating 
that tightly bolted plates add very materially to the conductance 
of a rail joint. Also recent tests of joint resistances on unbonded 
tracks which have been in service for a number of years show 

150211°— 19 5 

66 Technologic Papers of the Bureau of Standards 

that only about 5 per cent of unbonded joints have a resistance 
less than 1000 feet of rail. 

AYhile these tests might in some cases account for the apparent 
increase in resistance of bonded joints, it is undoubtedly true 
that a large number of the mechanically applied bonds slowly 
increase in resistance and may or may not reach a stage, within 
the life of the joint, where corrosion becomes so serious as to 
require the replacement of the bond. 

While contact deterioration has been attributed to the differ- 
ence in the coefficients of expansion of copper and steel and to 
other uncontrollable causes, it is undoubtedly true that by far 
the most prevalent cause of such deterioration is natural and 
electrolytic corrosion resulting from the entrance of moisture 
between the two metals. Considerable evidence is at hand to 
show that, as a rule, on compressed-terminal bonds moisture 
finds admission to the bond terminal between the head of the 
bond and the steel on the bond side of the rail. Engineers of 
the American Railways Co. state that they have removed hun- 
dreds of compressed-terminal bonds after being in service for 
a time and that on nearly every one the corrosion had started on 
the shoulder of the terminal on the bond side of the rail. 

Experiments conducted by bond manufacturers have demon- 
strated that, under the action of a compressor, a bond terminal 
will begin to expand at the end opposite the head of the bond, 
and will gradually fill the hole toward the head as the pressure 
is increased. It is evident, therefore, that such failures as those 
reported by the American Railways Co. are the result of incom- 
plete compression and emphasize the necessity of careful atten- 
tion to this feature. 

The life of poorly compressed bonds is possibly lengthened 
by grinding or otherwise cleaning, at the time of installation, 
the web of the rail with which the bond terminal comes in contact. 
Company 10 reports that they have greatly increased the life 
ot their compressed-terminal bonds by this operation and attrib- 
ute it to the good contact between the head of the bond and the 
web of the rail, which they believe delays the entrance of the 
moisture to the terminal proper. 

Rail Joints and Bonds 67 

A few complaints have been registered against the pin-ter- 
minal bond on account of the steel pin being subject to corrosion 
when allowed to come in contact with the earth. It is claimed 
that corrosion of the steel pin takes place and rapidly rusts it 
out, thereby relieving the compression on the copper terminal. 

The great advantages of the pin-terminal bond as claimed 
by the friends of that type are, first, that it can be installed with 
uniform and consistent results by ordinary labor, while the 
compressed-terminal type requires careful and expert labor for 
satisfactory results; and, second, it can be installed without 
interruption to traffic or the danger of a derailment, which is 
possible when using a compressor across the rail. 

Regarding the first point there appears to be a division of 
opinion, some claiming that the compressed-terminal type is 
more nearly "fool proof" than the pin-terminal bond. It is 
claimed by advocates of the compressed type that the big New 
York companies, who are having such phenomenal success with 
the pin-terminal bond, were forced to the adoption of that type 
by company rules which forbid the use of a compressor or any 
other tool which might cause a derailment; and the rigid specifi- 
cations and extreme refinements which they have adopted in con- 
nection with the purchase and application of their bonds inclines 
one to the conclusion that this, rather than the first-mentioned 
reason, was the determining factor in making their selection. 

It is significant that the principal advocates of the pin-terminal 
bond are to be found among the larger operating companies. 
The Philadelphia Rapid Transit Co. and the Bay State Street 
Railway Co., both operating extensive systems, may be added 
to the list of New York companies already mentioned. The 
observations of J. B. Taylor, engineer of way for the Philadelphia 
Rapid Transit Co., may throw some light on this point. 

Mr. Taylor states that in a system of the size of that in Phila- 
delphia a number of repair jobs are usually in progress at the 
same time and it would not be practicable to have expert bonding 
men at every job at the proper time. When the rails are ready 
the bonds must be applied, and as pin-terminal bonds can be 
installed quickly and with a fair degree of uniformity by an 
ordinary trackman, they are found to be better suited for this 

68 Technologic Papers of the Bureau of Standards 

class of work than the compressed-terminal bond, which should 
not be installed by any but an experienced and careful workman. 

The Bay State Street Railway Co. has a thousand miles of 
track in and around Boston. They adopted pin-terminal bonds 
five years ago because of the ease and uniformity with which 
they can be applied and for their "fool-proof" qualities. All 
bonds are purchased under rigid specifications based on their 
own drawings. Many of them, consequently, are not standard 
products of the manufacturers. Types and sizes are selected 
by laying out on the drawing board the rail and joint-plate 
sections and then prescribing a bond that has plenty of clearance. 

The extreme care with which the New York Central & Hudson 
River Railroad Co. installs their pin-terminal bonds has already 
been referred to. The reported price of 35 cents per bond for 
labor on installation is an indication of the grade of labor and the 
care employed. It is said that bonds are installed in only freshly 
drilled or reamed holes, and that the bond terminals are cleaned 
and polished before they are expanded. A driving fit must be 
secured and if a hole is found to be larger than the bond terminal 
it is reamed out and a larger terminal inserted, or if the difference 
is small a larger expanding pin is used. Finally, the expanding 
mandrel and pin must be of the correct diameter to insure proper 
expansion. Too large a pin will tear the metal, while too small a 
pin will not insure complete expansion. Three companies are 
manufacturing the standard 16-inch 500000 cir. mil stranded 
bond employed by this road, which is a special product, as no 
other operating companies use the same bond. 

That the careful and expensive methods of bonding employed 
by this company are fully justified is indicated by the extremely 
small percentage of failures recorded on page 36 of this paper. 
The following set of specifications which are required by the Bay 
State Street Railway Co. are similar to those required by other 
large companies employing pin-terminal bonds and may prove 
of interest to bond purchasers : 


Definition of terms. — The word "company" where occurring in this specification 
shall mean the purchaser of the material hereinafter referred to, or its duly authorized 

Rail Joints and Bonds 69 

The word "contractor" where occurring in this specification shall mean the party- 
accepting the order to furnish the material hereinafter referred to, or its duly author- 
ized representative. 

General description. — The materials required under this specification are 4/0 A. W. 
gauge capacity bonds, for bonding around track joints. 

The completed bonds and the materials of which they are made shall conform 
in design and dimensions to the company 's standard drawings, hereby made a part of 
this specification and to the following requirements and tests: 

Conductor. — All bonds shall consist of the required number of annealed copper 
wires or ribbons, free from splints, flaws, or other defects, and having an aggregate 
cross sectional area, when measured/ at right angles to the axes of the individual 
wires, at least equal to that of 4/0 American wire gauge. 

Each of the individual wires or ribbons shall have a conductivity of not less than 
98X per cent of standard annealed copper at 20 C. 

Where stranded bonds are required the wires shall be concentrically stranded 
together in spiral layers having at least one complete turn in each 5 inches of con- 

The copper wires shall not vary more than 1 per cent from the nominal diameter. 

The copper ribbons shall not vary from the nominal widths and thickness more 
than the amount shown in the following table: 

Thickness, Variation, Width, in Variation, 

in inches in inches inches in inches 

o. 010-0. 050 o. 001 o. 10-0. 250 o. 003 

Terminals. — Where copper terminals are required they shall be in effect a unit 
with the conductor. This may be accomplished by upsetting the head from a portion 
of the conductor or by welding drop-forged terminals on the conductor, in which case 
the union between terminals and the cable shall be a clean weld, free from oxide. 

The terminals shall be of uniform size and shape, free from cracks, burrs, fins, 
slivers, and hard spots, and any machining on terminals shall be followed by careful 

The surface of the terminals as called for on the drawing shall be milled smooth 
or otherwise finished, the resulting surface to be strictly equivalent to that obtained 
by careful milling. 

Where steel terminals are required they shall be made of steel of good quality, 
soft, and carefully shaped to the dimensions specified, and shall be thoroughly tinned 
inside and out before soldering to the conductor. 

The soldered joints between the terminals and cable shall be carefully made with 
half-and-half solder and shall be free from imperfections of adhesion, excess of solder, 
or any other defects. 

Tests — Union between conductor and terminals. — All bonds with copper terminals 
may be tested as follows to determine the character of the union between the head 
and conductor: 

The stud of the bond shall be sawed lengthwise into four equal segments, allowing 
the saw to cut to but not into the conductor: 

These segments shall then be bent back, tending to separate the welded parts. 
If a clean, bright fracture is exhibited, with a surface entirely free from dark oxide, 
the weld shall be considered satisfactory. It is not essential that the lines of the 
individual wires or ribbons be entirely obliterated. 

70 Technologic Papers of the Bureau of Standards 

Flexibility. — The test for flexibility hereinafter described is not made a condition 
of acceptance, but may be made at the option of the company and accorded due weight 
in the determination of the relative excellence of the bonds submitted. This test 
shall be made by holding rigidly one terminal of a bond while the other end is given 
a longitudinal movement of three-sixteenths of an inch, and a transverse movement 
of three-sixteenths of an inch, and continuing the movement until the first ribbon 
or wire breaks. 

Inspection. — Samples shall be selected at random from each type and kind of bond 
received for inspection and determination if they comply with the specification; 
the samples shall consist of two bonds from each ioo and at least two bonds if there 
are less than ioo bonds. 

Rejection. — If 10 per cent of the selected samples fails to comply with the require- 
ments of the specification all bonds represented by these samples may be rejected 
and returned at the expense of the contractor. 

Method of shipment. — All bonds shall be so packed for shipment that they will be 
suitably protected from injury, each package being plainly marked with the number, 
type, and length of bonds, and the number of the company's order upon which ship- 
ment was made. 

The argument that it is easier to obtain uniform results with 
the pin-terminal than with the compressed-terminal bond is 
largely based on the assumption that it is easier to drive a pin into 
a bond terminal than to properly adjust and manipulate a com- 
pressor. This is no doubt true, but it is also a fact that workmen 
frequently drive the pins in crooked and thereby fail to get a 
uniform expansion. 

In the installation of compressed-terminal bonds not only must 
the compressor be properly adjusted so as to get an even bearing 
on the bond terminal but the maximum compression must be 
obtained in order to secure the best results. It is also important 
to keep the point of the compressor in good condition, and the axis 
of the screw should be at right angles to the opposite face of the 
compressor. The difficulty of knowing when complete compres- 
sion has been obtained has led some of the operating companies 
to the practice of testing each bond at the time of installation. 
If it does not come up to the proper standard the compressor is 
applied again or a new bond installed. 

One of the manufacturing companies is attempting to provide 
for this insurance automatically by building a compressor which 
will shear out a button of sheet metal when the proper pressure 
has been reached. Workmen will then be required to turn in a 
button for each terminal compressed at the close of each day. 

Rail Joints and Bonds 71 

Granting that attention to these details is obtained only by- 
conscientious and experienced workmen, it is also true that a greater 
variation in the size of holes is permissible with the compressed- 
terminal than with the pin-terminal bond. The great care exer- 
cised in getting holes of exactly the right size and the rigid speci- 
fications regarding the diameter of pin-terminal bond studs is 
necessary on account of the small amount of expansion that can 
be obtained in this type of bond. On the other hand, although 
not good practice, pressure can be applied with a screw or hydraulic 
compressor until a comparatively loose bond terminal has been 
made to fill the hole. 

Summing up the arguments we are inclined to believe that 
honors are about even respecting the two types, and we are led 
to the following conclusions: 

Excellent results are obtained with the compressed-terminal 
bonds when they are carefully installed by experienced men. 
Inexperienced, untrustworthy, or careless workmen should not be 
employed for their installation. The compressor should be kept 
in good condition in order that complete and even compression 
may be obtained. The testing of bonds immediately after instal- 
lation is also a good practice. Failure to comply with these 
requirements is likely to result in poor contacts, followed by 
corrosion and rapid deterioration. 

Results equally as good as those obtained under the best condi- 
tions with the compressed-terminal bond may be secured with 
the pin-terminal type when careful attention is given to the details 
of installation, particularly to obtaining a driving fit for the bond 
terminal before driving the pin. This condition can be controlled 
within narrow limits by specifying only machined-terminal bonds 
and giving close attention to the grinding of drills. The reaming 
of all holes with a straight reamer is a good practice and will add 
greatly to the uniformity of results. Moderate and fairly uniform 
results can be obtained with the pin-terminal bonds when installed 
by ordinary inexperienced laborers if they are provided with uni- 
form bonds and pins as well as drills which have been ground at 
the shop. This is owing to the fact that under these conditions 
the personal element has been largely eUminated, and so long as 

72 Technologic Papers of the Bureau of Standards 

the pin is driven home there is a fair assurance that a good and 
permanent contact exists. 

Compressed- terminal bonds are often excluded from use on 
rapid- transit lines where the compressor over the rail would offer 
a hazard to the safe passage of trains, also from special work 
where the sharp angles of the frogs sometimes prevent its use. 

Pin-terminal bonds are excluded from use on elevated roads 
and other tracks where wooden or steel guard rails prevent the 
use of a hammer in driving the pins. The steel pins in pin- 
terminal bonds are subject to corrosion and should be used with 
caution where they are likely to be subjected to excessive moisture. 

The following code of instructions for the installation of pin- 
terminal bonds is given by Howard H. George in the Electric 
Railway Journal of September 19, 1914. It represents the best 
modern practice and has been incorporated in Prof. Richey's 
Electric Railway Handbook. All of the 13 rules, with the excep- 
tion of 10 and 11, apply equally well to compressed-terminal 


i. Every roadmaster and foreman should see that one or more men in each gang 
are taught the proper way of installing bonds, and should be sure that any bonding 
done thereafter is performed by these men. 

2. When renewing rail or joint plates on single track in operation, care should be 
taken not to open or disconnect both rails at the same time, as this would open the 
return circuit by which the current returns from the cars to the power house. When 
it is absolutely necessary to open both rails, a long copper jumper should be installed 
to connect the open ends so that the path of the return circuit shall not be interrupted. 
This applies more particularly to road ends and interurban lines. 

3. Whenever any track is opened up and any ground wires for electric lights, 
lightning arresters, or other electrical apparatus which should be connected to the 
rail are found disconnected, they should be reported at once to the bond inspector, 
or distribution department, so that they may be repaired before the track is closed 
up. This is very important and should receive careful attention. 

4. No bond holes should be drilled until just before the bonds are ready to be 
put in. There are, of course, times when it is desirous to have the holes drilled 
before the rail is placed on the ties. When this occurs, it is necessary to place a 
tight-fitting plug in the hole as soon as it is drilled to avoid any possible introduction 
of moisture. To drill a hole a day or two before and not protect it from moisture 
means a film of rust in the hole, which will greatly increase the resistance of the joint. 

5. Old bonds should never be used again, because they become battered up in 
driving them out. Then, when they are put in again they will not make good con- 
tact with the rail, which means a poor bond. Where a bond is removed from the rail 
it is not advisable to use the same hole in putting in a new bond, unless some pre- 

Rail Joints and Bonds 73 

cautionary methods are used. The proper way is to drill a new hole, but as this is 
not allowable in some types of rails, ream out the old hole and use a bond with a 
special large size terminal. 

6. Great care should be taken with the drills used in making bond holes. If an 
improperly ground drill is used the hole will be irregular and oval shaped, thus 
giving a poor contact between the terminal and the rail. All dull and broken drills 
should be carefully boxed, labeled, and sent to the shop to be reground, where the 
company has installed a special machine for the purpose to do the work perfectly 
and at much less expense than could possibly be done by hand. 

7. In drilling bond holes never use oil to lubricate the drills. It is better not to 
use anything, but where it is absolutely necessary to use a lubricant, nothing more 
than a soda solution should be employed. 

8. Holes, after being drilled, should be carefully cleaned of any chips and wiped 
dry of any solution that may have been used to lubricate the drills. The holes must 
have a smooth and dry surface, so that the bond terminal will make a good contact 
all around. 

9. With a proper size hole, the bond terminal will make a very snug fit, not 
small enough to have to be driven with a heavy maul nor large enough to be put in 
easily with the hands. It should require a couple of taps with a hammer weighing 
about 3 pounds. With a heavy hammer or spike maul the head of the bond terminal 
is very likely to be battered and the taper punch struck on the slant, causing it to 
split and bend the terminal. 

10. After the bond terminals are in position, always drive the long steel taper 
punch entirely through the terminal, taking care to strike the punch squarely on the 
head. The small end of this punch should be dipped in some kind of heavy grease, 
such as track grease, just before it is driven through each terminal. The grease will 
lubricate the sides of the punch, thereby expanding the terminals and not drawing 
the copper with the punch. 

11. Drive into each of the expanded terminals one of the short drift pins, thus 
expanding the copper a little more. This pin should be driven in until it is just 
flush with the head of the bond terminal. 

12. The bond should then be shaped by straightening out the bond conductors, 
and forming them so that they will not be cut by either the track bolts or the splice 
bars. If it is a 36-inch bond, it should be so shaped that it will in no way interfere 
with the removal of the splice bars. 

13. The bond, and particularly the bond terminals on both sides of the rail, are 
to be painted with some good weatherproof paint, care being taken to see that the 
paint fills the space back of the terminal heads. 

(b) Stranded v. Ribbon Bonds. — Referring to the tabulation 
under question 3, on page 29, we find that a total of 579 200 
stranded concealed bonds were reported, as against 214 200 ribbon 
bonds. Considering that the section of the ribbon bonds is much 
better proportioned for the space usually provided for concealed 
bonds, this apparent preference for the stranded type may seem 
surprising. Manufacturers who have conducted laboratory life 
tests on the two types differ as to their relative abilities to with- 

74 Technologic Papers of the Bureau of Standards 

stand vibrations. The American Steel & Wire Co. states in their 
general catalogue that the stranded bond will remain intact longer 
than the ribbon bond, while the Electric Railway Improvement 
Co. affirms that many vibration tests on short head bonds have 
demonstrated that the ribbon bond will outlast the wire type. 
This seems to be another case where theories and laboratory 
experiments offer little evidence as to what will happen under the 
peculiar exigencies of service. The fact is that the majority, 
though not all, of the operating companies are using and prefer 
the stranded bond for concealed work and say that it is giving 
better satisfaction than the ribbon type and is not so sensitive to 
vibrations and the corrosive action of the joint plates. 

The secret of the matter no doubt lies in the fact that the con- 
ductors of the ribbon bond, not being twisted or wound together, 
are easily separated and isolated. The space provided for con- 
cealed bonds is wedge shaped, as seen in Fig. 15, and the movement 
of the plates tend to work the ribbons and strands upward under 
the fishing surface of the rail head. This is a common complaint, 
and has been mentioned by a number of engineers. The wires of 
the stranded bond, being twisted together, are not so likely to 
become separated and broken by this action. 

The effect of loose joints on head bonds installed in macadam 
or earth streets is similar to that on concealed ribbon bonds 
described above. Dirt works in between the ribbons, which are 
gradually separated and brought to the surface by the continual 
motion of the joint. This trouble has been recognized, and is now 
being largely overcome by the application of a clip around the body 
of the bond, which prevents the separation of the strands. This 
clip is illustrated in Fig. 17. 

(c) Use of Solder and Alloys with Mechanically Ap- 
plied Bonds. — The large number of bond failures in past years 
resulting from corrosion of mechanically applied terminals have 
led a number of companies to adopt the use of solder or a plastic 
mercury alloy as a third or intermediate metal between the copper 
and steel. In some cases compressed-terminal and twin-terminal 
bonds have been installed with solder, in other cases the terminals 
have been tinned either at the factory or on the ground before 
installation, while in still other installations both the steel of the 

Rail Joints and Bonds 75 

rail and the bond terminal have been amalgamated. While these 
various practices have found rather wide application and have 
many adherents, the manufacturers of bonds are a unit in their 
belief that the copper to steel contact can not be improved upon. 
They argue that the introduction of a third metal, having a spe- 
cifically high resistance, between the copper and steel will not only 
add to the contact resistance but might also be the source of a 
chemical action which will hasten rather than delay corrosion. 
To the argument that a third and soft metal is needed to take 
up the difference in expansion between copper and steel, they 
reply that the process of expanding or compressing a copper ter- 
minal so hardens it as to give it sufficient elasticity to take up this 
slight difference in expansion itself. In the absence of experi- 
mental data to substantiate these theories we are again forced to 
base our conclusions upon the best modern practice resulting from 
years of experience and upon the opinions of prominent engi- 
neers. We will consider the practices of soldering, tinning, and 
amalgamating in the order in which they are named. 

The several methods of soldering compressed-terminal bonds 
which have been used are well described in the following extract 
from a letter received from the Ohio Brass Co., which developed 
the thermobonding process, in answer to an inquiry requesting 
information on the subject: 

The compressed-terminal type of rail bonds has had the most general use in the 
past, due principally to the fact that it can be installed in a satisfactory manner with 
the comparatively low grade of labor that must be relied upon for work of this kind. 
One of the chief objections to the compressed-terminal type of rail bond has been the 
rather small contact area between the terminal of the bond and the rail. In case the 
bond is not properly compressed, the contact surface would corrode, further reducing 
the efficiency of the bond. In order to overcome this difficulty many railroads make 
a practice of soldering the head of the bond to the rail after it has been compressed. 
With this method it is customary to tin the bond terminals before they are compressed 
and after the compression to heat them with an ordinary blow torch, applying solder, 
so as to form a perfect contact between the head of the bonds and the rail, thus supple- 
menting the contact secured by the compression and excluding the moisture from the 
plug portion of the bonds, at the same time giving an electrical contact which is not 
liable to deteriorate. The soldering process, however, adds considerable to the 
expense of the installation. 

The O-B thermobonding process was developed with a view of securing the advan- 
tageous results of soldering compressed-terminal bonds, at the same time furnishing a 
simple and cheap method of making this application. The charge of thermite is set off 
on the opposite side of the rail from the bond head and generates sufficient heat so 

76 Technologic Papers of the Bureau of Standards 

that the bond head can be soldered to the web of the rail. Many of the largest roads 
in the country using compressed-terminal bonds with solder changed over to the thermo 
process as soon as it was put on the market, and a great many bonds have been installed 
in that manner. However, it is an added refinement which is not considered essential, 
and as it adds considerable to the expense of the installation it has not had a universal 
use. The process requires some care in the installation, and for this reason, where a 
low grade of labor is used, further difficulties are encountered. 

Where a railway wishes to secure a very high grade of bonding and is willing to take 
the pains to use the proper care in installing the bonds by the thermo process, it is a 
very excellent method and is quite successful. 

The thermobonding process here described has fallen into dis- 
repute, and disuse so that at the present time it is employed very 
little, if at all. This has been the result, not only of the causes 
mentioned in the above quotation, but because of the injury done 
to the bond terminal and the web of the rail by the excessive 
heat generated by the thermite. We quote Company 16, on whose 
tracks a few bonds were soldered for demonstration purposes : 

After three years the bonds were removed for inspection, but we are sorry to say 
that they were in bad condition. The terminals were black and not soldered to the 
rail at all in the hole, and the excessive heat from the thermite had burnt the rail, 
which made it brittle and caused the steel to rust and depreciate. 

Conditions similar to these were also reported by A. P. Way, 
electrical engineer for the American Railways Co. Company 15 
says regarding the thermobonding process : 

About four years ago a thermosoldering process developed by the Ohio Brass Co. 
was made a part of the standard process of installing a bond. It is supposed to make 
a complete union between the copper bond and the rail. In most cases it appears to 
do so, but there have been cases where such bonds were removed several months 
after installation and the contact between bond and rail has been black instead of 
bright, thus showing poor contact. This may have been due to poor workmanship in 
the installation, particularly since bonds installed by this process test good after 

A test made by the Chattanooga Railway & Light Co. in November, 1910, shows the 
resistance of the thermo process bond contact to be approximately one-half the resist- 
ance of the compressed type without soldering. 

In addition to Companies 14 and 15, which have employed the 
thermobonding process, may be mentioned Companies 7, 20, and 
26, which solder their stud and twin terminal bonds by the aid of 
a blow torch. 

The following letter from Company 20, describing the method 
as practiced in Tacoma, Wash., will be found of interest: 

We are sending you, parcel post, one of our standard 250 M c. m. twin- terminal 
bonds. The American Steel & Wire Co. have made a special die for the bonds they 

Rail Joints and Bonds 77 

furnish us. You will notice that the face of the terminal which comes in contact with 
the ball of the rail is a flat smooth surface. In applying these we are very careful to 
see that the ball of the rail is chipped smooth and deep enough so that all rust spots 
are cut out. Although the terminals are tinned, we redip them immediately before 
they are applied. We also are careful to see that the holes in the ball of the rail and 
the chipped surface is perfectly clean and well tinned. The bond is driven with the 
rail hot and the solder fluid. After the bond is in place, solder is applied to the upper 
edge of the bond with an iron which makes a reinforcing fillet. 

Before adopting this type of bond as a standard, we applied a number of them 
without solder. Resistance measurements were taken immediately after the bond 
was applied and at intervals thereafter. We noticed in a number of cases that the 
resistance of the bonds increased, in some cases slightly and in some cases materially. 
Since we have been using solder in the application of this bond, we have made a large 
number of resistance measurements and have not noticed any change in the resist- 
ance of the bond. 

The bond here described is similar to the Form C twin-terminal 
bond, but having a broader face and a square shoulder to hold the 
fillet of solder. 

There seems to be a fairly general agreement that solder improves 
the contact of mechanically applied bonds, but that when applied 
by the thermo process the chance of burning both the bond ter- 
minal and the rail are so great as to offset the benefits that might 
accrue therefrom. In view of the excellent results which are 
being obtained with the compressed-terminal bond under the 
present improved methods of manufacture and installation, the 
additional expense of soldering this type appears to be hardly 
justifiable. The soldering process requires skilled labor, addi- 
tional time, and extra material. If half this time and expense be 
devoted to careful and improved methods of drilling holes and 
compressing the bond terminals, equally good results could be 
obtained. The use of solder in connection with twin-terminal 
bonds will be further discussed when considering that type. 

The process of tinning stud-terminal bonds before installing 
them has been employed by a number of companies, and appar- 
ently with general success. One of the strongest advocates of 
this practice is found in E- Hey den, superintendent of overhead 
construction for the Indianapolis Traction & Terminal Co., who 
is very positive in his belief regarding its value. Mr. Hey den 
states that he has used tinned-terminal bonds for years, and that 
in removing bonds from old rails he has invariably observed that 
corrosion has been much worse on bonds which had not been 

78 Technologic Papers of the Bureau of Standards 

dipped. He goes further and says that with a bond tester he is 
able to distinguish between dipped and undipped bonds when 
testing the resistance of rail joints. 

Tinned bonds are also being used by the Empire United Rail- 
ways, of Syracuse, where it is stated they show less corrosion than 
undipped bonds. 

The process of amalgamating bond terminals before compress- 
ing them is now finding favor with a number of operating com- 
panies, among which is Company 25, from whose report the fol- 
lowing quotation is taken: 

After using many types of bonds and rail joints, the writer has standardized on the 
use of compression bonds in connection with, which is used the so-called H. P. Brown 
pastic and solid alloys. These alloys have proved very valuable, in that they take 
care of any grooves which may be cut in the bond hole when drilling the rail for a 
bond, and also, being live materials, take care of the difference in coefficients of expan- 
sion as between copper and steel at the connection of the bond to the rail. These 
alloys also form a protective coating over the copper and steel, preventing corrosion 
near the point of contact. 

These alloys have also been used by the American Railways 
Co., which believes them to be valuable in excluding moisture 
from the bond terminal. 

These alloys are useful only where they are confined, as they 
soon corrode and lose their effectiveness when exposed to the 
atmosphere. In the New York power stations, where they were 
used on copper switches to reduce the contact resistance, they 
were effective for about three months, after which time the 
switches showed a higher resistance than before the application 
of the alloys. 

Although the practice of tinning and amalgamating stud- 
terminal bonds has not been very extensively employed and com- 
paratively little information regarding it is available, it appears 
that in some installations they have proved very valuable. Both 
processes are quite inexpensive, and on installations where cor- 
rosion of terminals has been a chronic trouble they can, no doubt, 
be used to advantage. The utility of many materials and prac- 
tices of this nature depends largely upon the personal element of the 
workmen involved. One man may learn how to apply compressed 
terminal bonds and obtain good and uniform results by his indi- 
vidual methods. The same man may utterly fail to get results 

Rail Joints and Bonds 79 

with pin-terminal bonds or with soldered or amalgamated com- 
pressed-terminal bonds. Splendid results obtained with the use 
of any type of bond or material are usually the result of the indi- 
vidual efforts of some person who has mastered that particular 

(d) Mechanically Applied Head Bonds. — The twin- terminal 
bond and tubular-terminal, or O-B, type J bond, shown in Figs. 7 
and 8, are included under this heading and are so similar in con- 
struction and with reference to their features of installation that 
they may well be considered together. 

These types were developed as the result of a demand for a 
bond that could be installed without removing the joint plates and 
without undue interruption to traffic. Experience had demon- 
strated that long bonds which spanned the joint plates were sub- 
ject to theft, and the types here mentioned were made as short 
as possible in order to reduce this loss as well as for the sake of 
economy. The short length has naturally resulted in considerable 
breakage from vibration, particularly on loose joints, and this is 
perhaps the most prevalent cause of failure. 

While these types are best adapted to open track where they 
are not subjected to vehicle traffic they have been used to some 
extent in earth and macadam streets. The cost of renewing a 
concealed bond in city streets is shown by figures in Section III to 
range from $1 to $3, depending upon the nature of the pavement. 
It is argued that the cost of installing a short head bond is so 
small in comparison to this that a company is justified in its use, 
although the depreciation in city streets may be high. 

Very little testimony has been secured regarding the use of the 
tubular-terminal bond, as this type has been on the market but a 
short time. The extent to which it is being adopted on new roads, 
however, is not only evidence of the general demand for a short 
head bond but an indication of faith in this particular type. The 
following information is pertinent regarding its use. 

In answer to an inquiry on the subject the Northern Ohio Trac- 
tion & Light Co. says: 

Replying to your letter of the 2d instant, I wish to advise you that the O-B type J 
rail bond which we have used has given us very satisfactory service. 

80 Technologic Papers of the Bureau of Standards 

The Detroit Railway has the following comment to make : 

Replying to yours of the 2d instant, relative to our experience with the O-B type J 
rail bonds, I beg to state that we have used a great many of these bonds on suburban 
work and they seem to be working out very well. I do not recommend them for city 
street work or for places where vehicles can in any way strike them, as they, like all 
other bonds of this character, are liable to shear off from the rail. 

George F. Silvia, electrical superintendent for the Albany South- 
ern Railroad, states that a systematic rebonding of all the tracks 
on that road will soon be commenced and that the O-B type J 
bond will be used exclusively. He believes that where these bonds 
fail others should be installed by drilling new holes in the head of 
the rails, as it would not be practicable to attempt to drill out the 
old terminals. 

Recent technical press notices state that on the new Grand 
Rapids- Kalamazoo line, 50 miles in length, and in the construction 
of the Fort Wayne & Northwestern Railway the O-B type J bonds 
are being used. 

The testimony relative to twin-terminal bonds is more abundant. 
Company 11 employs 60 000 and Company 42, 33 000 of the total 
113 700 bonds of this type reported. On both roads its use is con- 
fined to open track. As these companies are among the largest 
users of twin-terminal bonds in the country their experience should 
be significant. Company 1 1 says in answer to question 6 : 

Twin terminal and compressed terminal bonds usually last as long as the track; 
the percentage of failure is very small. 

Company 42 reports that in the seven years of operation 7357 
bonds on an original installation of 33 000 have been replaced. Of 
these 800 had been stolen. This is equivalent to an annual failure 
of approximately 3 per cent. 

Good results have also been secured by the Cincinnati Traction 
Co., where a suburban line of 25 miles in length has been bonded 
with twin-terminal bonds. Practically no failures have been 
experienced and the bond is said to be easily installed by ordinary 
unskilled labor. In the May, 191 5, issue of Electric Traction is a 
description of the new Waterloo-Cedar Rapids line in Iowa; 4/0 
twin-terminal bonds are employed and were installed at the rate 
of 100 bonds per day by four men with electric drills. 

Rail Joints and Bonds 81 

In contrast to the complimentary experience of these companies 
is to be found the opinions of a number of engineers who for one 
reason or another have not been pleased with this bond. The Bay 
State Street Railway Co. reports that they have used some twin- 
terminal bonds but found them short lived on account of breaking 
at the terminals or loosening of the studs. 

Company 24 reports that of a trial installation in 19 10 of 1000 
twin-terminal bonds 75 per cent were stolen in 1913-14. With a 
number of these the studs, as well as the body of the bond, were 
removed showing that the contact was poor. 

The Virginia Railway & Power Co. installed a few of these bonds 
but was not entirely pleased with the results obtained. 

Several engineers have been consulted who object to drilling 
into the heads of rails, saying that it is likely to weaken them and 
increase the chance for cupping. This argument, however, does 
not appear to be founded on definite facts. 

That the contact resistance of twin-terminal bonds, as ordi- 
narily installed, increases gradually with time seems to be a well- 
established fact. Some of the answers to question 18 giving 
information on this point are here repeated. Company 7 says: 

A new twin terminal bond will equal about 3 feet of rail, when the old ones will 
average about a^A or 5 feet. 

Company ii. — We have found no difference in resistance between old and new 
bonding, so long as the bonds remain unbroken and their terminal contacts are 

This statement is made with reference to twin-terminal and 
compressed-terminal bonds. 

Company 26. — Tests of compressed-terminal bonds and twin-terminal bonds are 
practically the same, 4 to tf/i new, 5 to 6 when old. 

The letter from Company 20, quoted above, also states that the 
resistance of unsoldered twin-terminal bonds increases after 
installation, in some cases slightly and in some cases materially. 

The Bureau of Standards recently tested 40 joints on the Wash- 
ington, Baltimore & Annapolis Railroad near Washington. 
These joints were bonded with 4/0 twin-terminal bonds which 
had been in service about seven years on 80-pound rails. The 
test was made on 3 feet of joint, and the highest and lowest resist- 

150211°— 19 6 

82 Technologic Papers of the- Bureau of Standards 

ances were 9.9 feet and 6.0 feet of adjacent rail, respectively, the 
average being 6.90 feet, or 0.0000828 ohm. 

A test made by the Bureau of Standards on a single-rail joint, 
newly bonded by the American Steel & Wire Co., with one 4/0 
twin- terminal bond, showed a 3 -foot joint with plates bolted in 
place to have a resistance of 0.0000349 ohm or 2.91 feet of 80-pound 
rail and with plates removed to have a resistance of 0.0000745 
ohm, which is equivalent to 6.2 feet of the adjacent 80-pound rail. 
While this latter figure is not materially less than the resistance 
of the average old joint it is much lower than that of individual 
joints. The joint plates undoubtedly add somewhat to the con- 
ductance of joints even on old installations but just how much it 
is difficult to say. The tests conducted by the Bureau of Stand- 
ards on unbonded joints which have been referred to before indi- 
cate that on old joints the function of the plates as far as aiding 
the return circuit is concerned is practically nil. It is altogether 
possible that observed changes in the resistance of bonded joints 
have frequently been attributed to deterioration in contacts 
when as a matter of fact they have been largely the result of loosen- 
ing and rusting of the joint plates. 

Considering all of the information at hand it appears to be 
more than probable that twin-terminal bonds as well as other 
mechanically applied bonds gradually increase in resistance with 
time. This increase in resistance with the twin- terminal type, 
though small in some cases, becomes quite appreciable in others, 
and on the average remains within the limit of good practice for 
a period of years, on joints which have been carefully installed. 

That the use of solder in connection with these bonds will fore- 
stall this contact depreciation is undoubtedly true. Its adop- 
tion, however, should depend upon local conditions and the per- 
sonnel of the force of workmen. 

Summing up the features of the mechanically applied head 
bond we find the following to be applicable to both types here 

These bonds are short and therefore comparatively inexpen- 
sive. They can be rapidly installed with very little interruption 
to traffic, the total cost of installation, including bond, being about 
50 cents each on new work. When used on city streets they can 

Rail Joints and Bonds 83 

be installed without removing joint plates, but on this type of 
construction they are subjected to vehicle traffic, which is likely to 
shear them off of the rail. On open track they are subject to theft, 
but this loss is much smaller than with longer bonds. It can be 
reduced by painting with black paint, thus rendering the bonds 
less conspicuous. In some cases they have been protected by 
iron plates bolted to the joint. Owing to the shortness of the 
bond, failures frequently occur from breaking of the strands, par- 
ticularly on poorly maintained joints. The contact between cop- 
per and steel slowly, though as a rule not seriously, depreciates. 
This may be prevented by soldering and to some extent by tinning 
the terminals. 

(e) Electric- WEU> Bonds. — The practically universal demand 
for something better than a soldered contact and a substitute for 
the purely mechanical contact has been responsible for the wide 
adoption of the electric-weld bond within the past few years. The 
720 000 of these bonds reported by the operating companies is 
an indication of the extensive use which it has found during the 
comparatively short period of its manufacture. Although it has 
been used for the most part as a head bond it is now coming into 
use more and more for concealed application to the web of the rail. 

There seems to be practically no question regarding the per- 
manency of the contact that these bonds make with the rail, but 
some criticism has been directed against some of their other 
features, particularly to the breaking of the ribbons and to the 
inconvenience of using the bonding car on tracks over which 
traffic must be maintained. The question has also been raised as 
to what effect if any the welding heat has on the steel of the rail. 
The following quotations from written reports and conversations, 
it is hoped, will throw some light on these questions. 

The New York State Railways, of Syracuse, are using large 
numbers of electric-weld bonds where they are believed to be the 
best bond available. The engineers state that they can be 
installed in paved streets b)^ removing a couple of bricks and the 
expense is not over 60 cents per bond, while the replacement of 
a concealed bond on similar construction would cost about $3. 
When installed in macadam streets a wire is twisted about the 
middle of the bond to prevent the ribbons from separating and 

84 Technologic Papers of the Bureau of Standards 

working to the surface. The need of this precaution has been 
recognized by the manufacturers, who now provide bonds with a 
clip as shown in Fig. 17. The company has been using the EA 
type and has experienced a rather high percentage of failures 
from breaking of the ribbons. Recently, however, the ET type 
has been adopted from which much better results are expected. 
These two types are shown in Figs. 4 and 5 and have been pre- 
viously discussed. 

The New York Railways, of Rochester, are using practically 
nothing but electric-weld bonds. On new work where there is 
heavy traffic a concealed bond is welded to the web of the rail and 
an EA and an EB type to the head, making three bonds per joint. 
Very little trouble from traffic is experienced in paved streets, but 
in macadam and earth streets a number of bonds have been 
broken by vehicles. This company has also recently substituted 
the ET for the EA type. 

The Hudson & Manhattan Railroad Co. is using 600 000 cir. 
mils electric-weld bonds and finds them satisfactory for new work. 
They are not so satisfactory for maintenance as the bonding car 
can not be derailed in the tunnels, and the expense of operating 
it at night for small repair jobs is excessive. This company claims 
that if the carbon electrodes are kept well back on the bond terminal 
and not allowed to fuse the ribbons near the bend in the bond that 
good results may be obtained with the EA type. 

Company 14. — After a great deal of experience here and elsewhere with various 
types of bonds, we have come to the conclusion that an electric- weld bond of con- 
siderable length to give the bond the proper flexibility is the only type of bond to use. 

The experience of Company 13, which has been quoted under 
answers to question 6, is similar to that of other companies in that 
the ET bond has been substituted for the EA type and better 
results are expected from it. 

Unfortunately, this new type has not been in service a great while 
and the companies are not prepared to make definite statements 
regarding it. The fact, however, that no complaints have been 
heard concerning it is good evidence that it is an improvement 
upon the EA type and will show a much smaller percentage of 

Rail Joints and Bonds 85 

Regarding injury to the rail by heat there seems to be very 
little definite information. A number of engineers have ex- 
pressed a question or fear regarding this point, but with possibly 
one or two exceptions no company has definitely reported any 
broken rails from this cause. The Cleveland Railway Co. states 
that several rails have cracked where bonds have been copper 
welded to the web of the rail. 

The Los Angeles Railway, which has over 300 000 electric-weld 

bonds, writes the following letter in reply to an inquiry on this 

question : 

In reply to your letter of the 3d instant regarding broken rails as a result of heating 
the web in applying bonds, wish to advise that in the several thousand brazed bonds 
we have installed under the plates or concealed on the web we have never had a broken 
rail caused by heating the web. 

It is reported that some fractures have occurred in rails through 
holes which had been punched in the web, but that such failures 
are easily prevented by reaming out the holes, thus relieving the 
strain around them. 

In the absence of more definite complaint on this score it is 
safe to say that the injury to rails resulting from the heat generated 
in welding rail bonds is so small as to be negligible and can be 
practically disregarded in the selection of a type of bond. 

The matter of interruption to traffic is something that will 
depend upon local conditions and will have to be met by each 
company in a manner depending upon a variety of circumstances. 
A number of companies operate their bonding car at night on 
tracks from which it is difficult to divert the day traffic. Upon 
suburban tracks, for maintenance work, it is usually possible to 
sidetrack the bonding car, or if not it may be derailed to accommo- 
date infrequent traffic. 

The following is a summary of the properties and features of the 
electric- weld bond : 

The bond is made for either head or concealed application. 
Either type can be installed on new work for from 50 cents to 60 
cents. The head bond is short and has experienced a rather 
high percentage of failures from the breaking of the ribbons, as 
well as from ignorance and carelessness in its application. Mod- 
ern improvements are materially reducing these failures. 

86 Technologic Papers of the Bureau of Standards 

The bond makes a very low-resistance contact with the rail, 
which does not depreciate with time. 

The shortness of the bond and the strength of the contact has 
made the theft of this type, on interurban lines, far less than that 
of other types. 

The bond is used successfully in paved and other types of city 
streets, where it can be cheaply installed. Being subjected to 
vehicle traffic, however, occasional failures must be expected. 
Where used in earth or macadam streets a clip should be employed 
to prevent the ribbons from working to the street surface. 

The installation of this bond is accomplished with the aid of a 
bonding car, the first cost of which is not justified on small prop- 

Maintenance bonding can not be accomplished without inter- 
ruption of traffic or considerable inconvenience in derailing the 
car or operating it at night. 

Some fear has been expressed regarding the heating of the rail 
webs, but the failures from this cause have been so infrequent that 
no great importance should be attached to them. 

The following account of the apparatus and methods used in 
connection with the electric-weld and the copper-weld rail bonds 
was furnished by the Electric Railway Improvement Co. Some 
of the apparatus referred to is shown in Figs. 6 and 6a. 


On direct-current lines electric- weld rail bonds are installed with a small car meas- 
uring 6 feet io inches in length and about 5 feet 10 inches in width. The frame of 
the car is of structural steel covered with an oak floor and carried on four 20-inch 

The electrical apparatus consists of a rotary converter and transformer with the nec- 
essary switches, circuit breakers, controller, resistances, etc., for its safe and con- 
venient operation. 

The rotary converter is provided with a clutch and is used as a motor for the propul- 
sion of the car along the track. 

The bonding clamps for electric welding are located at both sides at one end of the 
car over the rail s and have adjusting screws with hand wheels for bringing the same 
into position for service. 

To avoid interference with traffic, a screw jack with bevel-gears termination in cranks 
at each side of the car, is fixed under the center of the car frame. By means of this 
jack the car can be raised for the purpose of turning and rolling from the track to avoid 
interference with traffic. Depending on conditions, the car can be removed from the 
track in from one to one and a half minutes and may be replaced on the track in a 
similar length of time. 

Rail Joints and Bonds 87 

The entire car is covered with a canopy top which carries the trolley pole. 

On alternating-current lines the car is not necessary, and the AC voltage is carried 
directly from the trolley to a portable transformer. The welding clamp is attached 
to the transformer, which is provided with wheels for rolling along the track. This 
apparatus may be lifted from the track in a few seconds. 

On high-tension lines the trolley voltage would pass through a step-down trans- 
former before entering the portable. 

For electric welding, a current of from 25 to 50 amperes is drawn from the trolley; 
this varies, depending on the trolley voltage and the size of bond to be attached. For 
welding a 4/0 rail bond to the rail, an alternating current of about 2500 amperes at 5 
volts is employed, which is obtained on a direct-current line by converting and trans- 
forming about 25 amperes at 550 volts taken from the trolley. To make a weld, the 
current is applied for a period of from 45 seconds to 2 minutes, depending on condi- 

An average of 100 4/0 bonds per 10 hours may be installed with the car operated 
by a bonder and two helpers. For best results the bonder should be a man of average 
intelligence, while the two helpers may be laborers. 

The welding apparatus is not sold by the manufacturer, but is put out under a lease. 
The clauses of most importance in the lease are those stipulating that during the life 
of the patents all rail bonds installed with the apparatus shall be purchased from the 
manufacturer at prices given in schedules attached to the lease, that the prices for 
bonds as given in the schedules are guaranteed by the manufacturer and the only 
change in the prices is due to the fluctuation of the market price of lake ingot copper 
in the bonds, that the manufacturer will furnish an expert to instruct the railway 
company's men in the proper operation of the apparatus, and at the expiration of the 
patents, under which the apparatus is leased, the manufacturer will transfer title in 
the apparatus to the lessee. 


The bonding cars as put out by the manufacturer carry a melting furnace suspended 
from the rear of the car for copper welding. This type of welding is especially adapted 
to the installation of large conductors around special work, and for attaching feeder 
cables, etc., to the rails. Bonds and cables of any capacity may be attached by this 
method. It is also used for joining third rails both electrically and mechanically 

Copper welding is effected by pouring molten copper. A mold of suitable refractory 
material is employed, the size of the same depending on the section of the conductor 
and the size of the terminal or contact area desired. The end of the conductor lies 
in the terminal mold. A short channel connects the terminal mold proper with 
another chamber or reservoir. The rail having been properly cleaned at the point 
of weld, the mold with the bond wires is clamped in position. The molten copper, 
on being poured into the mold, impinges on the ends of the bond wires and the steel 
at the point of weld, and flows on into the reservoir, which finally becomes filled; 
continuing to pour the molten copper, it backs up into and fills the terminal mold. 
The excess of copper poured over that required to form the bond terminal proper 
has the effect of raising the temperature of the steel within the area of the bond ter- 
minal, to a proper welding heat; the steel then readily unites with the molten copper 

88 Technologic Papers of the Bureau of Standards 

The mold is then removed and the block of copper formed by the reservoir is taken 
off by cutting with a chisel the small section of copper which connects it with the 
bond terminal. 

Copper welding may be used to electrically and mechanically join third rails. 

The copper-welding outfit is also furnished independent of the bonding car. This 
particular apparatus is operated with kerosene oil, the crucible containing the 
copper being placed in a furnace similar to the one used in the coke outfits. 

The copper-weld furnace car is equipped with from three to five kerosene furnaces 
and is designed for use on large installations. 

The copper-welding apparatus is put out under a lease similar to that used in the 
case of the electric-weld cars. 

This method of bonding, while being adapted for installing large cables, etc., is 
especially suited for bonding steam-road electrifications and new electric lines where 
the electric power is not available at the time of installation. For this type of work 
it is not necessary to have any of the apparatus on the tracks while operating. In 
fact, the single kerosene-furnace outfit may be carried along by two men, and need 
never occupy the track. 

A modification of the Electric Railway Improvement Co.'s bond- 
ing car has been devised by the Cleveland Railway Co., and em- 
ployed by them with considerable success, particularly for their 
special work bonding. In this process, which is described in the 
Electric Railway Journal of August 7, 19 15, silver solder is used 
for a brazing metal instead of brass, and the union of the bond 
with the steel is therefore affected at a much lower temperature 
than is necessary with the standard equipment. It is claimed 
that the lower temperature and a different arrangement of elec- 
trodes enable the bond to be applied with a much smaller current 
than is used by the regular bonding car. This smaller current is 
obtained by a small rotary converter which can be mounted on 
a motor truck and connected to a step-down transformer by 
flexible conductors. The transformer is carried about from joint 
to joint over an entire intersection and with no interruption to 
traffic. It is claimed in Cleveland that this modified equipment 
and method has made possible the bonding of special work on a 
daylight schedule and with no broken rails or interruption of 

( / ) Oxy- Acetylene- Welded Bonds. — This type of bonding is 
accomplished by welding bonds with forged or cast terminals to 
the rail, usually the head, and employing pure or fluxed copper 
to build up the terminal of the bond along the rail head. So far, 
it has not found a very wide application, although it has been 
used in Minneapolis and St. Paul for a number of years with 

Bureau of Standards Technologic Paper No. o2 

Fig. il 

Apparatus for applying bonds with the oxy- 
acetylene flame 

FlG. 19. — Apparatus for applying joint plates ivith 
the electric arc 

Rail Joints and Bonds 89 

marked success. It possesses practically all of the advantages of 
the electric- weld bond and is free from the objection of requiring 
an expensive equipment and of interfering with traffic. Although 
it does not make quite as strong a contact with the rail as does 
the electric- weld bond, the contact is permanent and does not 
deteriorate with time. No injuries to rails have been reported 
from the use of the flame which apparently does not produce a 
higher temperature in the rail than is generated by the electric 

The Minneapolis Street Railway Co. has the following to say with 
respect to their practice of bonding : 

The only type of bonds used during the past three or four years is a 250 000 cir. 
mils, U shaped, T head, bond welded on the side of the head of the rail, and in 
special cases on the side of the flange of the rail or on the base of the rail, using 
acetylene torch for welding. We use this type of bond on all types of construction 
from the light rail in earth streets to 100-pound guard rail on special work in paved 
streets, and on heavy T rail construction laid on concrete base with block, asphalt, 
or stone paving. We have at present between 14000 and 15000 such welded 
bonds in service. 

Unskilled labor is used for bonding work; that is, the laborers are instructed in the 
proper handling of the tools and are then competent to do the work. It is found 
unnecessary to do any cleaning of the steel before welding bonds. For acetylene 
welding we use compressed acetylene in acetone tanks and compressed oxygen in 
tanks, with torch specially developed for this class of work. Pure copper is used for 
welding and no flux is required. 

Complete cost of installing welded bonds of 250 000 cir. mils capacity in large 
numbers is between 50 and 60 cents per bond including cost of tools and machinery. 

The failure of this method of bonding to have found more 
extensive use is apparently not the result of any defects or dis- 
advantages connected with the process, but is more than likely 
the result of other causes. The recent concerted action of the 
manufacturers of both bonds and acetylene in providing and 
advertising proper bonds and material for this method of bonding 
will no doubt act as a great stimulus to its further adoption. 

The following statements regarding this method of bonding rail 
joints have been abstracted from information submitted by the 
Ohio Brass Co. The apparatus referred to is shown in Fig. 18. 


For applying rail bonds the equipment consists of the necessary tanks of com- 
pressed oxygen and dissolved acetylene, which may be purchased by the railway 
companies direct from the manufacturers, who have warehouses and factories quite 
generally distributed over the country, thus insuring prompt and convenient service. 

90 Technologic Papers of the Bureau of Standards 

A pressure regulator and set of gauges are provided for both the oxygen and acety- 
lene tanks. These are changed from tank to tank as they become empty, the empty 
tanks being returned to the manufacturers of the gas. A hose extends from each 
set of pressure regulators and is connected to a blowpipe or torch by means of which 
the gases are mixed and adjusted to give the proper condition of the flame. 

A grinder for cleaning the rails will also be required and may be either of the elec- 
trical or hand operated type. A pair of colored goggles to protect the eyes of the 
operator will be found advisable for constant work, although the rays of light from 
the oxy-acetylene flame do not affect the eyes and skin as do the rays from the 
electric arc. 

A small truck provided with handles will be found convenient for conveying the 
tanks of gas in service and can be purchased from the manufacturers or built from 
blue prints furnished by them. 

Clamps for holding the bond to the rail while welding on one terminal will be sup- 
plied by the manufacturers, as will also the necessary flux wire. The latter is a metal 
especially prepared for the purpose of attaching the bond to the rail and for building 
up the head of the bond. It is three-sixteenths of an inch in diameter and is fur- 
nished in coils of about 50 pounds, which the customer should cut into suitable lengths. 

For the proper application of bonds the work of only three or four men is required; a 
working foreman who watches and assists the blowpipe operator; blowpipe operator; 
and one or two grinders, the number depending on whether an electric or hand grinder 
is used. The grinders have time for other necessary duties as well. 

The bonds require on an average about 3 cubic feet, at atmospheric pressure, each 
of oxygen and acetylene, and the actual time required to apply a bond is from four to 
five minutes. 

(g) Bonding of Manganese And Other Special Work. — The 
recent use of manganese steel for special work which is too hard to 
drill has led to some difficulties in bonding. On this point Com- 
pany 3, in speaking of electric- weld bonds, says: 

This method of bonding manganese special work and steam-railroad crossings has 
been the only one that we have had any success with so far, on account of not being 
able to drill the manganese steel for compressed type. Some of the steel companies 
have undertaken to insert a soft-steel plug in the rail which can be reamed out to re- 
ceive the bond terminal, but we find that the contact between the plug and steel is in- 
sufficient to carry current for any time. 

Other companies report no such trouble from inserts in manga- 
nese steel, and the failures in this particular instance may have 
been due to the fact that the plugs were inserted in the rails by 
the aid of an oxyacetylene flame rather than having been cast in 
as is customary. 

Company 27. — We bond around all important special work where any quantity of 
current is to be provided for by welding (with an oxyacetylene flame) heavy copper 
lugs to the web or base of the rail joint. To these lugs are attached stranded weather- 
proof covered cables of from 500 000 to 2 000 000 cir. mils, cross section, which spans the 
entire special work. 

Rail Joints and Bonds 91 

Question E-W-84 in the May issue of the A. E. R. A. asks what 
is the best method of bonding manganese special work and railroad 
crossings other than by using the plug holes provided. The fol- 
lowing answers to this question as found in the June and July num- 
bers are given below: 

Chas. E. Fritts, Electrical Engineer, Metropolitan Street Railway Co., 
Kansas City, Mo. — We use copper cables underneath the special work and single 
bonds by brazing. 

Edw. J. Blair, Electrical Engineer, Metropolitan West Side Elevated 
Railway Co., Chicago, III. — The best way of bonding manganese special work and 
railroad crossings is to jump around them with copper-cable conductors, making the 
connection to ordinary rails behind the manganese special work. 

A. V. Brown, Engineer Maintenance op Way, Lake Shore Electric Railway 
Co., Sandusky, Ohio. — Run cable around manganese special work. 

H. P. Bell, Electrical Engineer, San Francisco-Oakland Terminal Rail- 
ways Co., Oakland, Cal. — In bonding special work and railroad crossings, we have 
found, after much investigating, testing, and experience, that it is cheaper and more 
efficient to bond around crossings and special work (unless electrically continuous and 
easily drilled) with such size cables as the current density and voltage drop allowed 
demand at that particular location. We place such cable or cables in trunking filled 
with tar compound and bury them about 6 inches below the tie, connecting them to 
the through rails by means of seven-eighths-inch separate compression-bond terminals. 
Thus the track special work may be repaired or replaced without necessary repairs to 

C. L. Cadle, Electrical Engineer, New York State Railways, Syracuse, 
N. Y. — The New York State railways' practice has been for the last seven years to 
install welded bonds on all track joints needing electrical connections. From our 
experience the welded bond adheres to manganese rail and special work as well as 
open-hearth and Bessemer steel. At railroad crossings in addition to bonding the 
joints each crossing is jumpered by placing a piece of 4/0 copper cable, or larger, 
around the crossing, so that in case the bonds on the joints of the special work are 
broken off the current will be carried through the jumper cable. 

C. D. Emmons, General Manager, Chicago, South Bend & Northern Indi- 
ana Railway Co., South Bend, Ind. — We use brazed bonds for all of our work. 

H. G. Throop, Superintendent Line and Buildings, New York State Rail- 
ways, Syracuse, N. Y. — I believe that the best method of bonding manganese 
special work is to use welded bonds to the head of the rail, double bonding, using a 
10-inch and 18-inch bond. The special work should also be cabled, i. e., one 500 000 
cm. cable installed for each track, attaching the ends of this cable to two 4/0 cross 
bonds welded to the rail at a point back of the special work each way from the 

H. F. Merker, Engineer Maintenance op Way, East St. Louis & Suburban 
Railway Co., East St. Louis, III. — On account of the fact that special work with 
heavy cables carrying current around it may be considered dead for all the time 
except when a car is passing over it, there have been various methods of bonding used. 
In fact, some engineers have gone so far as to leave all bonds off of such special work. 
We know of no better way to bond special work than to leave a soft plug for drilling a 

92 Technologic Papers of the Bureau of Standards 

bond hole, but this hole may be placed at any convenient part of the special work 
piece and need not be in line with the bolt holes or near them, as all that is really 
necessary is one tap connecting it with the return circuit to keep the switch piece 
alive, and it is not necessary that each joint be bonded. 

E. H. Schofield, Engineer Power and Equipment, Minneapolis Street 
Railway Co., Minneapolis, Minn. — In addition to cables spanning special work 
and bonded to rails, bonds may be welded by acetylene torch to the side of the head 
or guard of special work parts. 

H. A. Clarke, General Manager, Ithaca Traction Corporation, Ithaca, 
N. Y. — My experience has been that it is the best practice to bond around such spe- 
cial work, using bond terminals of whatever type may be in use, soldered to sufficient 
length of oooo copper wire. 

J. B. Tinnon, Engineer Maintenance op Way, Chicago & Joliet Electric 
Railway Co., Joliet, III. — I know of no better method of bonding manganese steel 
special work than by using the plugs in the castings. It is almost impossible to weld 
any kind of bond to the manganese casting and the heating of the casting would no 
doubt tend to destroy the value of the manganese steel. It is also nearly impossible 
to drill the casting, so I think that the method of using the plug holes and bonding to 
a through cable is the best method that has yet been developed. 

H. E. Gough, Engineer, Elmira Water, Light & Railroad Co., Elmira, N.Y. — 
We double bond all connections in special work, drilling the bond holes on the ground. 
Better connections are insured if drilling is done on the ground rather than in the 
shop. It is also generally wise to provide supplementary bonding around special 
work where the traffic is heavy. 

Geo. H. Pegram, Chief Engineer, Interborough Rapid Transit Co., New 
York, N. Y. — We find it most convenient and satisfactory to have all bonded holes 
provided by the manufacturers with copper plugs for signal bonding. We do not use 
manganese rail for negative return, as we have an opportunity to make use of the open- 
hearth steel guard rail for that purpose. The guard rail is bonded to manganese rail 
by means of clipped bonds. 

In a paper presented at the annual convention of the South- 
western Electric and Gas Association, Galveston, Tex., May, 19 15, 
G. W. Smith, engineer, San Antonio Traction Co., San Antonio, 
Tex., said: 

Our experience with rail bonds as ordinarily applied has not been such that we 
felt justified in depending on them to carry the current around the special work. We 
tried to get information as to whether or not copper cables could be welded to the 
steel rail by the thermite process, but in so far as we were able to find out this had 
never been done, so we proceeded to do some experimenting. The result was that 
we succeeded in making a weld which gives a contact area equal to or greater than 
the cross-section area of the rail. The cable used, which is 800 000 cir. mils, is welded 
to the lower side of the flange of the rail, and a section cut through the weld shows a 
shading in color from steel at the top to copper color at the bottom of the weld. The 
cable enters the weld at the bottom and is therefore in contact directly with the cop- 
per film at the bottom of the weld, which insures minimum contact resistance. We 
are using this method of bonding on all reconstruction work. 

Rail Joints and Bonds 93 

The opinions here expressed are overwhelmingly in favor of 
bonding around special work. In several answers no mention is 
made of bonding the individual members of the special work. 
Although they are called upon to carry current only when a car 
is passing over them, it is obviously good practice to provide for at 
least light bonding to the main return circuit. 

Although the consensus of opinion seems to be in favor of weld- 
ing or brazing bonds to manganese steel, a number of engineers 
are apparently having good results from the bonds installed in 
the soft plugs provided by the steel companies. Only one answer 
refers to the possibility of injuring the manganese steel by weld- 
ing bonds to it. Experience, however, seems to be sufficiently 
complete to show that any fear regarding such injury is unwar- 

(h) Bonding of Converted Steam Roads. — The bonding of 
steam railroads in preparation for electrification presents some 
problems somewhat different from those obtaining on the ordinary 
city or suburban track. The conditions which are, as a rule, 
similar, are briefly as follows: (1) Joint plates are in place and 
the removal of them would not only be expensive but unsafe with 
the passing of high-speed trains; (2) the location is usually such 
as to invite theft of exposed bonds; (3) safety is of prime impor- 
tance, and the adoption of any bonding method from which a 
broken rail might possibly result could not be permitted. 

These conditions have forced several railroads to the adoption 
of a special type of bond which, so far, has been confined to this 
class of work. The objection to the removal of joint plates ex- 
cludes the concealed type of bonds and necessitates the use of 
either a short bond or one long enough to span the plates. Objec- 
tion has been made to the short head bonds both on account of 
the breakage of the strands and ribbons under heavy traffic as 
well as to the alleged injury done the rail by the welding heat 
and the drilling of the head. As the adoption of a long exposed 
bond would mean continual loss and trouble from theft, a long 
compressed or pin terminal bond has been devised and adopted 
by several railroads which can be inserted under the joint plate 
and locked in position by the removal of a single bolt. This bond 

94 Technologic Papers of the Bureau of Standards 

is provided with one terminal attached and one detached, and a 
sharp crimp or loop is put in the body of the bond near the end 
containing the fixed terminal. One of the end bolts of the joint 
plate is removed and the bond is threaded under the plate from 
this end. When the bond is in its proper position the bolt is 
replaced, which engages with the loop and locks the bond. The 
loose terminal is then soldered to the end of the bond and expan- 
sion or compression of the terminals is then affected in the ordinary 
manner. It is stated that in the electrification of the Chicago, 
Milwaukee •& St. Paul Railroad this type of bond will be used 
and that it was adopted only after an exhaustive study of the 
subject. The Pennsylvania Railroad and the New York, New 
Haven & Hartford Railroad are also using this or similar types. 
(i) Double v. Single Bonding. — The proper size and number 
of rail bonds per joint appears to be a subject regarding which 
very little definite knowledge exists, and the present practices of 
the operating companies seem to be based upon arbitrary and 
irrational rules rather than upon theoretical considerations. 1 
W. A. Del Mar, in a letter published in the Electrical World of 
April i, 1909, comments upon the absence of information and 
standard practice upon this point, and gives the following four 
rules as having been used by the operating companies to deter- 
mine the capacity of bonds : 

(1) Making the conductor equal in capacity to the rail. 

(2) Asking advice of manufacturers. 

(3) Doing what other people have done. 

(4) Guessing. 

Realizing the irrational character of these rules, Mr. Del Mar 
attempted to determine the proper capacity of bonds by the 
following three more logical considerations : 

(1) Determine the magnitude of the continuous and maximum 
currents in the rail and make the bond large enough to prevent 
undue heating. 

(2) Make the bond large enough to give at least 90 per cent 
efficiency to the return circuit; that is, make the conductance of 
the bonded rails at least 90 per cent of the conductance of a 

1 The Bureau of vStandards is now making preparations to conduct experiments on the carrying capacity 
of different types of bonds under various conditions. The results of these experiments will be published 
at a later date. 

Rail Joints and Bonds 95 

theoretically continuous rail. The following equation is given 
by Mr. Del Mar for the efficiency of the return circuit : 

Eff . = Ll 


where K= average efficiency of bond, or the ratio of the con- 
ductance of the bond to the conductance of an equal length of 

L x = length of rail, 

L 2 = length of bond between terminals. 

(3) Make the bond large enough to conform to proper 
mechanical conditions. 

The author of the letter states that he was immediately frus- 
trated in his attempts by having no data on the capacity of bonds, 
and after a thorough search through trade literature and scientific 
abstracts he came to the conclusion that no work was available as 
to the amount of current that short bonds would carry under 
various conditions without undue heating. He suggests this as 
a splendid field for research by the colleges and commercial lab- 
oratories and scores the manufacturers for not having data on 
the capacity of their various bond products. 

That this subject is of immediate and practical interest, and 
one on which a great diversity of opinion exists, is shown by the 
following answers submitted to question E-W-88, asking for the 
practice of the companies regarding single and double bonding, 
published in the June and July issues of the A. E. R. A. : 

C. D. Emmons, General Manager, Chicago, South Bend & Northern Indiana 
Railway Co., South Bend, Ind. — We single bond our track in all cases excepting 
in city streets where leading to power stations. 

H. F. Merker, Engineer Maintenance oe Way, East St. Louis & Suburban 
Railway Co., East St. Louis, III. — Where s the return current is large and where a 
bad electrical joint would be serious, or where paving conditions make the opening 
of joints a serious matter, there is no doubt that double bonding is to be preferred. 
Remember the old maxim, " Do not carry your eggs all in one basket. " 

John Leisenring, Signal Engineer, Illinois Traction System, Springfield, 
III. — This company has adopted the practice of double bonding all track in pave- 
ment or other forms of streets. The tracks on private right of way are only single 

96 Technologic Papers of the Bureau of Standards 

H. P. Bell, Electrical Engineer, San Francisco-Oakland Terminal Rail- 
ways, Oakland, Cal. — We practice both double and single bonding of rails, depend- 
ing upon the current density, allowable voltage drop, size (conductivity) of rail, and 
capacity of bond. 

Charles E. Fritts, Electrical Engineer, Metropolitan Street Railway 
Co., Kansas City, Mo. — Double bonding. 

A. H. Babcock, Consulting Engineer, Southern Pacific Co., San Francisco, 
Cal. — Tracks are both single bonded and double bonded, according as the load re- 
quirements change with varying localities. No hard and fast rule can be laid down, 
but bonding should be done always with reference to the potential gradient in the rail. 

A. V. Brown, Engineer Maintenance of Way, Lake Shore Electric Rail- 
way Co., Sandusky, Ohio. — Double bond all tracks in pavements. 

Edward J. Blair, Electrical Engineer, Metropolitan West Side Elevated 
Railway Co., Chicago, III. — it is our practice to single bond our tracks, but this is 
rather a matter for local conditions to determine. Whenever the resistance of the 
return circuit can be kept within bounds by single bonding, it should be done. 

C. L. Cadle, Electrical Engineer, New York State Railways, Syracuse, 
N. Y. — The practice of this company has been to bond each joint with at least one 
4/0 bond. At locations where the current density is such that the size of this bond 
will not carry the current imposed on it, as high as five bonds are installed at each 
joint to take care of the additional current. 

H. A. Clarke, General Manager, Ithaca Traction Corporation, Ithaca, 
N. Y. — I would state it is the practice of this company to single bond our tracks. 

George L. Wilson, Engineer op Maintenance op Way, Minneapolis, Minn. — 
The Twin City Lines use only single bonds. It has never been the practice of this 
company to double bond its tracks. 

H. G. Throop, Superintendent Line and Buildings, New York State Rail- 
ways, Syracuse, N. Y. — It is a good practice to double bond tracks which are heavy 
carriers of return current to the power house from congested districts and also to double 
bond in congested districts. In both Utica and Syracuse this practice is followed 
throughout the central portions of the city and on lines to which the greatest amount 
of negative cables, which lead back to the power houses, are attached. This method 
also gives some insurance for good bonding, as of course the two bonds give greater 
life than a single bond. 

George H. Pegram, Chief Engineer, Interborough Rapid Transit Co., New 
York, N. Y. — We double bond our rail joints. 

H. E. Gough, Engineer, Elmira Water, Light & Railroad Co., Elmira, 
N. Y. — We double bond our tracks in the central portions of the city and where traffic 
is heavy. In outlying sections single bonding is used. We are using a pin-terminal 
bond for this purpose. 

J. B. Tinnon, Engineer Maintenance of Way, Chicago & Joliet Electric 
Railway Co., Joliet, III. — We single bond tracks except where the return flow is 
very high, and then we double bond. The practice of double bonding for the pur- 
pose of having one good bond in case the other fails is, I think, a waste of money, as 
the same condition that causes one bond to fail will also cause the other to fail in 
most cases. 

J. C. Donald, General Superintendent Asheville Power & Light Co., 
Asheville, N. C — Bonding both track rails with cross bonds located every 200 feet 
is considered good practice. 

Rail Joints and Bonds 97 

F. M. Richards, Electrical Engineer, Atlantic Shore Railways, Kenne- 
bunk, Me. — It has been our practice to single bond our tracks, installing cross bonds 
between rails every thousand feet. 

These answers throw absolutely no light on the question as to 
just what is the safe carrying capacity of rail bonds. That a 
short copper bond attached to heavy masses of cold steel can not 
attain a dangerously high temperature is obvious even when 
carrying currents of the magnitude found on heavily loaded 
tracks. The contact resistance of one terminal of a mechanically 
applied bond in good condition is in the order of 0.000005 ohm. 
When carrying a current of 500 amperes, which is greatly in excess 
of currents ordinarily found in rails, there would be a dissipation 
of only 1% watts, and with 1000 amperes a dissipation of 5 watts 
per terminal. This is, of course, in addition to the heat generated 
in the copper of the bond, but as far as the contact is concerned 
there seems to be no practical limit to its capacity to carry current. 

Parshall in England, in an article on " Earth returns for electric 
tramways," published in the Journal of the Institution of Elec- 
trical Engineers, April 28, 1898, stated that experience with pres- 
sure contacts in central station work had demonstrated that 100 
amperes per square inch was the safe limit, but suggests that 50 
and even 25 amperes per square inch would be found more advis- 
able for rail bonds. As these figures are exceeded in practically 
all installations in this country they can not be regarded in any 
way as a practical limit for current density. 

A reference to the above answers will show that a number of 
companies, including some in large cities, install only single bonds 
and that several employ double bonding only when necessary to 
reduce the potential gradient on the tracks. This would indicate 
that double bonding is not necessary solely from the standpoint of 
capacity, but that its adoption is demanded by other considera- 
tions. One of the usual causes for its use is to insure safety, and 
not place reliance on a single bond in a permanent track where 
the repair of a bond would mean tearing up the pavement. 

Under extreme conditions double bonding is justifiable solely 
from the economic standpoint, and the factors which determine 
these conditions may also be used to determine the economic 

150211°— 19 7 

98 Technologic Papers of the Bureau of Standards 

replacement of deteriorating bonds. These conditions can be 
determined when the constants of a given system are known and 
the cost of power and the average current in the rails are 

Let W = weight of rail per yard, 

/ = root-mean-square current in rail over a 24-hour period, 

p = cost of power per kilowatt hour in dollars, 

P = cost of installing a bond, 

n = number of years bond will last, 

L = reduction in joint resistance resulting from installation 

of bond, expressed in feet of adjacent rail, 
r = rate of interest paid on invested capital; 
then the resistance of the rail is very close to 0.001/ W ohm per 
foot and the annual saving of energy in dollars due to installing a 

bond would be P X °*°£ x -±- X 24 X 365 = °-<*** PL P 
W 1000 ^ ° ■ w 

The annuity 2 required to retire the investment, P, on a new 
bond at the end of n years, its period of usefulness, is 



where R = i H 


When the annual power loss without the bond exceeds this 
annuity, it is obvious that the installation of a new bond would 
be a matter of economy. The limiting condition would be 
obtained when the energy charge is equal to the annuity. Equat- 
ing these two we get: 0.00876 -^ = p(— J, from which any 

quantity may be obtained provided the others are known. If we 
employ the equation to determine at what current double bonding 
becomes economical, we have 

WP (/?-!) 

0.00876 Lp R n - 1 

In order to apply this equation with six independent variable 
factors it will be necessary to assume a set of values which would 
obtain under normal conditions. 

2 American Handbook for Electrical Engineers, p. 830. 

Rail Joints and Bonds 99 

Let it be required to determine whether one or two 4/0, 10-inch 
compressed-terminal bonds should be used on 100-pound rails 
being newly installed under the following conditions: 

p =cost of energy =$0.01 per kw. hr. 

P = cost of installing bond = .60 
n = life of bond = 12 years. 

r = rate of interest = 5 per cent 

The resistance of one 10-inch 4/0 bond, including contact 
resistance, is very close to 0.000055 ohm, and would probably 
average more. Two such bonds in parallel would have approxi- 
mately one-half of this, or 0.0000275 ohm, which is also the 
decrease in the resistance of the joint resulting from the installa- 
tion of the second bond. As a 100-pound rail has a resistance 
very close to 0.00001 ohm per foot, L would be 2.75 feet. 

Substituting these values in the above equation we find that I 2 
is equal to 15 660 or / = 125 amperes. 

The Bureau of Standards has examined numerous railway load 
curves and has found that the ratio of the root-mean-square 
current to the all-day average, ranges from 1.25 to 1.4. If we 
use the lower of these values, which is more applicable for the 
heavily loaded lines with which we are concerned in this discussion, 
we get 100 amperes as the all-day average value of the current. 
With a load factor of 40 per cent this would give 250 amperes 
per rail as the value at the peak period. 

While the values here assumed are normal in every respect, 
they are undoubtedly on the side tending to make the limiting 
current small. A lower cost of energy, a shorter life of the bond, 
and a higher cost of installing a bond will all tend to give a larger 
current where double bonding becomes economical. As the 
values assumed for these three variables are obviously near the 
limit in the other direction, it is difficult to conceive how the 
economy point would be reached at any all-day average current 
value much less than 100 amperes. 

While a current of 100 amperes all-day average is entirely 
possible and undoubtedly exists on the rails of many properties, it 
is far in excess of what good electrolysis conditions would dictate. 
One hundred amperes on a 100-pound rail would give a drop of 1 

ioo Technologic Papers of the Bureau of Standards 

volt per iooo feet, and as 0.3 to 0.4 of a volt per 1000 feet, average 
for the 24-hour period, is considered the limiting potential gradient, 
consistent with good electrolysis conditions, it is seen that the 
latter condition would limit the current in the rails long before 
economy would demand double bonding. 

It is true that double bonding will reduce the potential gradient 
in rails, and several of the engineers quoted above seem to con- 
sider this as a determining factor for this practice. Its influence 
in this respect is quite small, however, as will be seen from the 
following calculations : 

Consider 10-inch 4/0 concealed bonds on 100-pound rails as 
before. With 60-foot rails we have approximately 59 feet of 
rail in series with the bond and the resistance of the two are 
0.00059 an d 0.000055 ohm, respectively, or a total of 0.000645 
ohm. Adding a second bond would reduce the resistance by 
0.0000275 or to 0.0006175 ohm, which is 95.73 per cent of the 
resistance of the rail with one bond. With 30-foot rails the effect 
would be more marked, and the maximum effect on the potential 
gradient resulting from the addition as a second bond would be 
where a short-head bond is installed on a joint previously bonded 
with a long-cable bond around the joint plates. 

A 30-foot rail bonded with a 36-inch 4/0 compressed terminal 
bond will have a resistance made up as follows: 27 feet of rail, 
0.00027 ohm, 3 feet of 4/0 copper, 0.00015 ohm, two contacts, 
0.000013 ohm, or a total of 0.000433 ohm, of which 0.000163 onm 
is due to the bond. If now a 4/0 electric-weld bond having a 
resistance, including contacts, of 0.000045 ohm be applied to the 
head of the rail the long bond will be shunted by approximately 
2}4 feet of rail in series with the short bond or by 0.00007 ohm. 
The two bonds in parallel will have a resistance of 0.0000487 ohm, 
making the resistance of one rail length of circuit 0.0003187 ohm, 
which is 73.6 per cent of the resistance before the application of 
the second bond. With 80-pound rails, 60 feet in length the 
corresponding figure would be 87.3 per cent. 

These calculations will substantiate the statement that, except 
perhaps under extreme conditions, double bonding in lieu of 
single bonding has only a secondary effect upon the potential 
gradient in return circuits and therefore upon electrolysis condi- 

Rail Joints and Bonds 101 

tions. When the potential gradient therefore begins to approach 
the maximum allowable limit, or the limit set by good practice, 
other and more effective means should be employed to reduce it. 

In all of the above calculations the effect of the joint plates 
has been neglected. Their effect would be to increase the con- 
ductance of the joint and therefore reduce the necessity of double 

Considering only tracks in which the potential gradient does 
not greatly exceed the values considered safe from the stand- 
point of electrolysis we find that neither carrying capacity, econ- 
omy, nor voltage drop will justify the practice of double bonding 
under ordinary conditions. The only factor remaining, therefore, 
which might justify the use of two or more bonds is that of insur- 
ance against the total failure of a bonded joint. 

It is difficult to say to what extent the probability of a joint 
failure is reduced by the addition of a second bond. If the 
failures are the result of loose joints the second bond is not much 
of an insurance, while if failures are the result of imperfect work- 
manship in installation the value of the insurance is much greater. 
It is altogether possible that two bonds of different types could 
be used to advantage for this purpose rather than bonds of the 
same type. A long bond around the joint plates might act as an 
insurance against the failure of a concealed bond and a con- 
cealed bond might insure the joint against theft of the exposed 
type. In employing a second bond as an insurance against 
the total failure of a joint it, of course, acts to improve operating 
and electrolysis conditions as well, to say nothing of reducing the 
power loss in the return circuit. These advantages when con- 
sidered together will ordinarily justify the use of the second 
bond on new and permanent tracks which are being installed in 
paved streets, where inspection of bonds is difficult and repairs 
expensive. The additional expense of installing the second 
bond at the time the new track is being laid is relatively very 
small and, in general, will be returned in reduced maintenance 
costs. The practice of double bonding would, perhaps, not 
ordinarily be justified on open track where the inspection and 
repair of bonds is less expensive and where the traffic as a rule 
is lighter. 

102 Technologic Papers of the Bureau of Standards 

(j) Economic and Other Considerations for the Replace- 
mi: xt of Bonds. — The installation of a bond on a joint already 
bonded may be considered either as a replacement or as double 
bonding, and no sharp line of distinction between the two con- 
ceptions can be said to exist. The term "double bonding" has 
been used with reference to joints on which two bonds are origi- 
nally installed or where a second bond is installed to supplement 
a new or old bond in a practically perfect condition. The term 
"replacement" will be used with reference to bonds installed to 
supersede or supplement old bonds which have failed or which 
are in a state of deterioration. 

The value of the resistance which a deteriorating bond must 
reach before economy will justify its replacement can be deter- 
mined from the equation given on page 98 of this paper, in which 
the reduction in the joint resistance resulting from the installation 
of the new bond becomes the unknown quantity. 

Transforming the equation for this purpose we find that 

0.00876 Pp 

' *(is) 

The following concrete example will illustrate the application of 
this formula. Let us assume : 

W, the weight of rail per yard = 100 pounds. 

P, the cost of replacing a bond = $0.80. 

p, the cost of energy = 0.0050 per kw. hr. 

n, the life of the new bond =10 years. 

r, the rate of interest = 5 per cent. 

Let us also assume that the current is of such a value as to 
give a drop of 0.8 volt per 1,000 feet as an all-day average on a 
perfectly bonded rail. This is greatly in excess of the current 
permitted by good electrolysis conditions and is not ordinarily 
exceeded, even in regions where the problem of electrolysis does 
not exist. The resistance of a 100-pound rail is 0.01 ohm per 
1,000 feet, which would limit the average current to 80 amperes. 
Taking 1.3 as the ratio between the root mean square and the 
all-day average current we get for our equation 7 = 104 amperes. 
The replacement cost of a bond is usually greater than the cost 
on new work and is therefore taken at 80 cents. One-half cent 

Rail Joints and Bonds 103 

per kilowatt-hour for power may seem low, but as it is assumed 
to be the cost of energy which will be saved by the application 
of the bond, it would not be logical to load it with fixed charges 
and operating costs other than the fuel. Upon this basis it is 
high rather than low. 

A replacement bond is often installed on old track which is 
partially worn out and which may be entirely replaced within a 
few years. Ten years, therefore, is considered as a liberal life for 
the new bond. The srcap value of bonds is small and is neglected 
in this discussion. 

Substituting these values in the above equation we find that 
L = 13.4 feet. As the new bond itself will test equal to from 3 to 
6 feet of rail this means that the old bond will test equal to about 
18 feet of rail before a new one can be installed with economy. 
This is, of course, far beyond the point of deterioration which good 
practice has established for the replacement of bonds. Electrol- 
ysis and voltage conditions ordinarily, therefore, demand a better 
return circuit than economy itself can dictate. In fact, it is doubt- 
ful if economy alone in many circumstances will justify any but 
the cheapest and simplest type of bonding. The Bureau of Stand- 
ards 3 and numerous independent investigators have demonstrated 
beyond a doubt that the character of most electric roadbeds is 
such as to shunt a large fraction of the current from the rails even 
when well bonded. With poor bonding the increased gradient 
along the track tends further to increase the leakage current which 
might easily reach a large per cent of the total current, except in 
the immediate vicinity of the negative bus. Parshall, in an article 
previously referred to, makes the following statement: 

In tests recently carried out in a line some 8 miles long it was found, by cutting 
the track at the middle of the line and inserting an ampere meter, that some 60 per 
cent of the current was returning through the earth itself. Tests made as to the con- 
ductivity of the earth return showed as a whole that it was about one and a half that 
of the rails, bonds, and fishplates, which would indicate that on an average about $2, 
per cent of the current was leaving the rails. In other words, the voltage drop in the 
earth return was but two-thirds of what it would have been had the current been 
wholly in the rails. 

3 The Bureau of Standards is now engaged in investigating the resistance of different types of roadbeds 
under various weather conditions. Tests are being conducted on experimental tracks built for this purpose 
as well as upon city and suburban lines. A full account of this work will be published at a later date. 

104 Technologic Papers of the Bureau of Standards 

In this connection the experience of the Virginia Railway & 
Power Co. is of interest. They report that several years ago the 
bonds on an alternating-current line of several miles in length were 
failing rapidly as the result of incorrect methods and poor work- 
manship on the original installation. A complete rebonding of 
the tracks meant a heavy expense which the road at that time was 
not prepared to meet. A careful study of the situation was made 
and after ascertaining that bonds were being omitted on a number 
of modern European alternating-current lines it was decided to 
continue operation without rebonding and to keep a careful record 
of the energy consumption from year to year. After three years 
of this practice, during which time practically all of the original 
bonds had failed, the road is in successful operation and no increase 
in energy consumption chargeable to poor bonds has been noted. 

Such a condition as here described would, of course, be utterly 
impracticable on an ordinary direct-current system on account of 
the pernicious electrolysis conditions which would obtain, to say 
nothing of poor operating conditions. 

The one thing which calls for and demands good bonding is good 
electrolysis conditions. Without the incentive to guard against 
trouble from this source it is difficult to say to what degree of 
deterioration a company is justified in allowing its return circuit 
to descend. 

(k) Standards for Replacement. — The practices of the com- 
panies regarding the replacement of bonds, as recorded in Table i 
on page 32, indicate a wide variety of standards, ranging from 
$y£ feet of rail to 27 feet of rail as the limiting resistance of joints. 
Among the answers there was not a single suggestion as to the 
manner in which any particular standard was established. 

The selection of a standard for replacement should be governed 
almost wholly by local conditions. There are so many elements 
to be reckoned with that it is practically impossible to suggest a 
hard and fast rule that would be applicable under all circum- 
stances. Even if such a rule or formula could be developed it 
would contain so many variable factors and qualifying consider- 
ations as to make its application impracticable. 

Frequent objection has been made to the universally adopted 
practice of referring to the resistance of a bond in terms of a 

Rail Joints and Bonds 105 

length of adjacent rail. The practice is a convenient one but 
obviously irrational, as it gives no indication of the actual resist- 
ance of the bond unless modified by supplementary data. The 
length of the joint tested, the length of the bond, and the weight 
of the rail must be known before the resistance of the bond in 
ohms can be calculated. A logical standard of replacement should 
obviously include these several factors, which would, however, 
greatly complicate the simple rule now in general use. It has been 
suggested that the replacement resistance of a bond should be 
defined in terms of its increase in resistance over that of a similar 
bond newly and properly installed rather than in some arbitrary 
length of adjacent rail, but this also has its objections. The in- 
crease in the resistance of a bonded joint results from breaking of 
strands and ribbons, corrosion of terminals, and loosening and 
rusting of plates, thereby reducing their ability to aid the bond. 
With welded bonds the deterioration from corrosion of the con- 
tact is nil while with the long cable bonds its effect on the total 
resistance of the bond is far less than in short concealed or head 
bonds. A standard therefore based upon, say, a 100 per cent 
increase in resistance, would be as irrational and arbitrary as one 
based upon a given length of adjacent rail. Moreover, with most 
of the standard bond-testing instruments, giving the resistance of 
a joint in terms of adjacent rail it would be difficult to discontinue 
the practice of referring to bonds in these terms. 

A standard which is simple, definite, and workable, even though 
it does not always meet the demands based upon a rigid technical 
analysis of the subject, is much to be preferred to one which 
attempts to meet these demands but is clumsy, complicated, and 
impracticable. A standard of replacement based upon the resist- 
ance of a given length of adjacent rail possesses the advantages 
here enumerated and from the standpoint of practical consider- 
ations is not so irrational as may at first appear. After all, the 
factors which limit the resistance of bonded joints are, ordinarily, 
electrolysis and operating conditions, and both of these are de- 
termined by the potential gradient in the return circuits If, there- 
fore, we assume the same current density in rails of different 
weights, which is by no means a violent assumption, we are con- 
sistent in basing our standard on a given length of rail, inde- 

106 Technologic Papers of the Bureau of Standards 

pendent of its weight, thereby limiting the voltage drop across 
the joint rather than its resistance in ohms. 

As to what this standard should be is a matter that should be 
determined largely by local conditions. Different standards 
might well be employed by a single company to meet the condi- 
tions on different types of construction or in regions requiring 
different degrees of electrolysis protection. It is altogether prob- 
able that it would be found advisable in some cases to employ 
different standards on city and suburban tracks. Answers to 
question 5 would indicate that the limiting resistance for a 3 -foot 
joint of from 6 to 10 feet of adjacent rail would ordinarily fall with- 
in the bounds of good practice. Greater lengths might be em- 
ployed in regions where no trouble from electrolysis is likely to 
exist, but on city streets these values should not be exceeded. 


It has been previously stated that the information available 
relative to the life and character of different types of welded joints 
is far less complete than that upon the subject of rail bonds. Some 
of the reasons for this lack of reliable information are set forth in 
the following quotation from one of the prominent electric railway 
engineers of the country : 

In my experience I will say that I consider the information gotten from the fur- 
nished answers from questions as above to be unreliable. There are but very few 
cities in which the engineering department has been in existence or that the engineer 
has held his position long enough to be able to make any statement about worn-out 
rails or worn-out joints. The intensity of traffic and the weight of cars bears an im- 
portant relation to the life of rails and joints and the conditions vary in different cities, 
and have varied so from conditions of 10 years ago that very few men are capable of 
intelligently answering the questions. 

Also the change in the material of which rails are made has been so great that unless 
this is considered the information is unreliable. The first Bessemer rails of a girder 
type were rolled about 1884. These rails were structurally defective, and the rail 
mill did not know how to roll the girder rail. Girder rails at this period weighed 56 
pounds to the yard and were about \Y 2 inches high. By 1890 the girder rail had 
grown in height to 6 inches and some of the sections were such that they could be 
properly rolled, but many engineers were still using sections that could not get the 
proper treatment in rolling. About 1893 the 9-inch girder rail was coming into general 
use and girder rails commenced to be a success. The rails rolled from 1890 to about 1901 
were Bessemer made from a good grade ore and gave a long life. About 1904 the grade 
of ore formerly used for making Bessemer rails was getting extremely scarce, so a lower 
grade was used, and for the next two or three years the rails furnished street railroad 

Rail Joints and Bonds 107 

companies were not as good as formerly furnished, and at this time the weight of cars 
had been materially increased, so that track life was short. About 1909 the open- 
hearth rail came into general use and apparently gives a very much better rail than 
that furnished previously, although the open-hearth rail has not been in use long 
enough to give absolute results on its life and durability. 

The increase in carbon in the steel used in rails has materially affected the reliability 
of the welded joint in some locality. 

Realizing as we do that the conditions here described are true 
to a large degree, we do not feel justified in attempting a close 
analysis of facts or in drawing any but the most cursory and gen- 
eral conclusions from the data and information at hand. The 
several types of welded and special joints will be taken up in turn 
and a brief summary of the facts available will be presented. 
Conclusions will be drawn and recommendations made only where 
the facts seem to warrant such. 

(a) The Cast Weux — The compilation of figures under question 
1 on page 49 show that of the 268 052 electrically continouus joints 
reported, 149 716, or more than 50 per cent of the total, are cast 
welds. In consideration of the fact that the installation of this 
joint has been practically discontinued these figures are an indi- 
cation of an early popularity at a time when the welding of rail 
joints was looked upon by many as a rather bold experiment. 

The excellent results and long life obtained from this joint as 
reported by a number of companies, under questions 4 and 6, 
would seem to indicate that the discontinuance of its use in recent 
years is not the result of its inability to meet the conditions imposed 
upon it, but rather to the fact that other and more modern types 
of joints are meeting the requirements of service and installation 
in a more satisfactory manner. 

Several causes are given as responsible for the failures of the cast 
weld. A frequent cause of complaint is that a true weld is not 
effected and the rails loosen up in the joint. This not only aug- 
ments wear and cupping but adds materially to the resistance of 
the joint. Some operators are inclined to the view that the cup- 
ping of the rails at the joint is frequently due to the softening of 
the steel, which is said to result from the excessive heat of the 
molten metal. This is apparently the most serious fault that has 
been found with the cast weld and is no doubt largely responsible 
for the almost total abandonment of this type of weld which has 

10S Technologic Papers of the Bureau of Standards 

occurred in recent years. The large amount of molten metal used 
in the cast weld maintained the rail ends at a high temperature for 
a considerable length of time, and the slow rate of cooling had an 
annealing and softening influence which was later manifest in 
cupped and worn rail heads. 

None of the modern welding processes employ the large amount 
of metal that was used in the old cast weld, and the cooling is con- 
sequently much more rapid. Moreover, the greatest heat is now 
confined to the weh and base of the rail, and it is therefore 
extremely doubtful if the heads of modern steel rails are seriously 
injured by any of the welding processes in use at the present time. 
The cupping which so frequently develops in cast-weld joints 
may also have been augmented by a difference in the resilience 
between the joint and the adjacent rail or to imperfect surfacing, 
either of which would give rise to pounding and uneven wear on 
the rail. 

In 1908 Parshall, in an article already referred to, has the fol- 
lowing to say regarding the cast weld: 

Another method of somewhat the same nature as the process of welding is that known 
as the "cast weld, " or the " Falk joint. " This joint is made by pouring molten metal 
into a metal mold clamped round the rail joint. The surfaces of the cast metal that 
come in contact with the mold and with the rail joint are chilled, and are thus pre- 
vented from forming a perfect weld. I believe it has been asserted that a weld is 
effected. It seems, however, extremely doubtful, since without the use of a flux a 
weld is almost impossible between cold wrought steel and molten iron. The rail 
expands after the metal is poured around it, and remains expanded until after the 
cast iron has set, and finally resumes its former size. This affords a slight clearance 
for expansion and contraction, and accounts for the mechanical success of the joint, 
which, if carefully applied, makes when new a perfect mechanical track; although, 
in the writer's mind, the difference of resilience between the part surrounding the 
casting and the remaining part of the track may eventually cause uneven wearing 
away of the rail. 

The clearance above spoken of undoubtedly admits a certain amount of moisture, 
so that by the formation of oxide the resistance of the joint increases in the course of 
time. From the results of tests which I have at hand, it also appears that the electrical 
resistance of this joint, even when new, varies considerable^; so that, considering the 
low voltage restrictions in this country, it should be used in connection with an efficient 
form of bond. Owing to the rigidity of the joint, however, copper bonds will undoubt- 
edly be found more durable in conjunction with it than with a fishplate form of joint. 

(6) The Thermite Welded Joint. — The failures of the thermite 
weld and the rather large percentage of cupped joints which have 
been reported from time to time have been for the most part on 

Rail Joints and Bonds 109 

the old type of weld and should not be considered as evidence 
against the greatly modified and newer joint which is referred to on 
page 20. Very little evidence is at hand regarding this modified 
type of weld, but if it is successful in preventing the cupping of the 
joint one of the chief objections to the thermite weld will have been 
removed. In contrast to the old cast weld the thermite weld 
requires but a small amount of welding metal, and this metal 
actually unites with the steel of the rail to obliterate the joint. 
The small amount of metal and the perfect union contribute to 
give the joint a resilience practically the same as that of the adja- 
cent rail, and the wheel is thereby enabled to pass without encoun- 
tering a hard spot in the track. Moreover, the metal being 
continuous, the joint has a high conductance which does not 

Among the objections to the thermite weld has been recorded the 
facts that the process is comparatively slow and that in repair 
work where it is necessary to insert a short length of rail to take 
the place of a badly cupped railhead two distinct welds are neces- 
sary where other processes require only a modification of a single 
weld. These conditions are said to make the process less satis- 
factory on old rail than for new work. 

The San Antonio Traction Co. adopted the new thermite weld 
in the reconstruction of their tracks, which was begun in October, 
1 91 3. A full account of the work was given by G. W. Smith, 
engineer for the company, in a paper presented at the. annual 
convention of the Southwestern Electrical and Gas Association, 
at Galveston, Tex., in May, 191 5. Mr. Smith made the following 
remarks : 

We have made a total of 3000 welded joints on track laid with concrete roadbed, 
and since the tracks have been put in operation we have had two breaks in the rails. 
One of these occurred at a crossover in the fall of 19 14 and the other near a bridge in 
the winter of 1914. The first welds were made in the winter of 1913, and these have 
been through two winters and one summer. The joint which broke near the cross- 
over is in this lot. The joints which were put in in the summer of 1914 have been 
through one summer and one winter, and the joint near the bridge was in this lot. 
In so far as we are able to judge from our experience of the past 18 months, we are 
convinced that, from the mechanical as well as electrical standpoint, the best type 
of permanent construction is obtained by welding the joints in the rails and using 
steel ties in concrete. 

no Technologic Papers of the Bureau of Standards 

As the new type of thermite joint has been in operation for 
only about two years, it is too early to form definite opinions 
regarding it. That it is an improvement over the earlier types, 
however, there seems to be no question. Reports from other 
companies who have adopted it or who have installed it on an 
experimental basis will be looked forward to with interest. 

(c) Thb Electrically-Welded Joint. — The relatively small 
number of n 697 electrically-welded joints which were reported in 
answer to question 1 is no indication of the extent to which this type 
of joint is being used at the present time. Its use is confined for 
the most part to large properties where the Lorain Steel Co. is 
installing it by contract in great numbers. It is being used largely 
for reclaiming old track where cupped joints are repaired by 
using a "dutchman" and extra-long splice bars. Four welds are 
required on such joints instead of three, and the expense and time 
is about one-third greater than for an ordinary joint. The popu- 
larity of this type of joint is due largely to the rapidity with which 
it can be installed and the consequent small amount of inter- 
ference with traffic. The conductivity of the joint is 100 per 
cent or better, which is greatly to its advantage. 

The Boston Elevated Railway Co. has several hundred miles 
of new and old track welded by this process, for which Harry M. 
Steward, chief engineer of maintenance of way, has only words of 
praise. He declares this to be the cheapest method he has found 
for reclaiming old track which is cupped and in bad condition. 
He says, further, that breaks are more prevalent in old than in 
new rails, and attributes this to the fact that old rails have strains 
in them which are sometimes responsible for the fractures which 
are developed with the welding heat. These fractures usually 
occur through one of the holes of the rails, and it is said that if 
new rails with no drillings are welded that no breaks would occur. 
In fact, Mr. Steward says that the failures on their new rails 
have been practically nil. 

The Worcester Consolidated Street Railway Co. installed 7800 
electrically- welded joints in 1902, which have given entire satis- 
faction. Few failures have occurred, and the most of these have 
been fractures through bolt holes. 

Rail Joints and Bonds 1 1 1 

Company 10 has used the electrically- welded joint on old rails 
with cupped joints which they wish to maintain for six or seven 
years. They do not advocate the joint on new work, as they 
believe the spot welding introduces strains into the web of the 
rail which are likely to cause subsequent failures. 

A head-supporting splice bar has been developed and is used 
on the rail sections which have shown a tendency to fail from a 
depression of the head. This splice is shown in Fig. 10. 

Summing up the evidence at hand we find that the electric-weld 
joint has been used in the past principally for the reclaiming of 
old and partly worn rail, and there only where comparatively 
large contracts have made the use of the elaborate equipment 
practicable. The joints have a conductance equal to or greater 
than the solid rail and are welded rapidly. Badly cupped joints 
are welded by using a "dutchman," extra-long splice bars, and 
four instead of three welds. The local and intense heating of the 
web introduces strains into the rail which frequently are the cause 
of fractures and breaks. Such failures are less frequent on new 
than on old rails, and particularly less frequent on undrilled rails. 

The following information regarding this welding process has 
been prepared from statements submitted by the I/Orain Steel 


The electrical welding of rail joints by the Lorain Steel Co. 's process is done exclu- 
sively by them tinder contracts entered into with railway companies. The Lorain 
Steel Co. furnishes all apparatus, material, and labor for welding the joints. The 
railway company supplies the necessary current for operating the welding equipment 
and prepares the track ready for welding. Where repair work on old rail is to be done, 
this consists of removing the paving around the joints to the bottom of the rail, the 
ties not being disturbed; removing the old splice bars and bond wires, and bringing 
the rail ends to the proper surface and line. New rail is welded either before or after 
the paving is done, a space being temporarily left around the joint in the latter case. 
After the welding is completed the railway company replaces the paving. Where 
traffic is not heavy the work can be carried on continuously, day and night. On 
double track portable crossovers are made use of by the railway company for diverting 
the traffic. In sections where the traffic density is too great to permit of the use of 
crossovers the work is done between the hours of midnight and 4 or 5 a. m. The porta- 
bility of the welding equipment making it easy to weld in one place for a few hours 
and then move to some other locality. About 15 minutes are required to complete a 
joint. Three welds are made on the standard and head-support joints, and the cur- 
rent is on about 2 minutes for each weld. At 500 volts about 250 amperes are required, 

1 1 2 Technologic Papers of the Bureau of Standards 

or about 125 k\v. , this current being on for 6 minutes to each joint the power consump- 
tion amounts to about 12J2 kw. hrs. per joint. The parts to be welded are brought to 
a "w elding heat by means of the resistance offered by the materials to be welded to a 
large now of current under a low voltage. The current is used simply for heating and 
no arc is formed. The welding current is supplied at about 7 volts. After the proper 
degree of heat is attained the current is cut off and the parts are forced together under 
very heavy pressure which is held in place until the metal has cooled below the critical 
temperature of crystallization or recalescence. 

The process is applicable to all kinds of track construction where the motive power 
to be used is electricity. In welding open track on elevated railways or on surface 
lines on private right of way, expansion joints are made use of to provide for expan- 
sion and contraction. These are placed from 800 to 1000 feet apart and at the ends of 
all curves. 

A welding equipment consists of four cars, provided with railway motors, and is 
operated in three units. The first car contains a motor-driven air compressor and a 
sand-blast apparatus. With this the rails and bars, at the points where the welds are 
to be made, are entirely cleaned of dirt and rust. For the second operation of welding 
two cars coupled together are provided. The first of these cars carries the welding 
transformer and pressure apparatus suspended from a crane, in the car to the rear of 
this a rotary converter, inverted, changes the direct current from the trolley to an 
alternating current. A regulator maintains the welding voltage practically constant 
at 300 volts regardless of the fluctuations of the trolley voltage. A range of from 325 
volts to 650 volts direct current can be operated on. The current from the regulator 
is passed to the welding transformer, and is here stepped down to the welding voltage, 
of about 7 volts with about 25 000 amperes. The bars are placed over the joint, one 
on each side of the rail web, and the welding contacts are brought into place to engage 
the middle of the bars, and this weld is made first, after which one end of the bars is 
treated and then the other end. In this way the bars are in an elongated state when 
the ends are welded and on cooling off exert a powerful pull to bring the rails ends 
together, thus leaving practically no joint at all. In the car carrying the rotary con- 
verter a switchboard carries instruments for recording the voltage and amperage. In 
the welder car suitable water tanks and circulating system is provided for circulating 
water through the welder transformer and the contacts to keep them cool. The third 
operation consists of grinding the head of the rail to a true running surface, and the last 
car carries suitable grinding machines for this purpose. This car also carries a furnace 
for melting the spelter used in making the head-support joint. The cars have been 
so designed that they can be readily loaded on gondola freight cars and are shipped 
from city to city in this manner. 

Where it is necessary to ship the equipment by railroad the company requires 3000 
or more joints to make it justify them in taking a contract, but where the machines 
can be run over trolley tracks they accept work for as few as 500 joints. 

(d) The Arc-Weeded Joint. — The application of the electric 
arc to the welding of rail joints is comparatively recent, and 
although numerous companies have installed a few arc-welded 
joints as an experiment, many of these have not been in service 
long enough to afford reliable information as to their ability to 
meet the demands of service. Moreover, the arc has been used 

Rail Joints and Bonds 113 

in the construction of so many different types of joints that time 
alone will be able to determine their relative merits. 

Special plates as shown in Figs. 11 and 12 are manufactured 
by the Indianapolis Switch & Frog Co. and have undergone sev- 
eral changes and improvements during recent months. Formerly 
the ends of these plates were brought to points instead of being 
cut off as shown, and this brought two welding seams to a junc- 
ture at the base of the web. With a similar plate applied on the 
opposite side of the rail two more seams were brought to the same 
region. This condition resulted in numerous breaks through the 
base of the rail and led the company to adopt the modified form 
of plate here shown. Not only have the plates been cut off to 
prevent the juncture of the welding seams but one plate is made 
slightly higher than the other, and they are then staggered longi- 
tudinally. The result is that no two seams come directly opposite 
each other and the rail at no point is heated by more than one 
seam. It is claimed by the manufacturers that these modifica- 
tions have greatly reduced the possibility of failures from frac- 

The following information regarding this welding process has 
been prepared from statements submitted by the Indianapolis 
Switch & Frog Co.: 


A general view of the welding outfit as manufactured by the 
Indianapolis Switch & Frog Co. is shown in Fig. 19. This welder, 
which consists largely of resistance coils to reduce the potential 
at the arc to about 70 volts, is mounted on light wheels similar to 
those of a wagon, weighs 1700 pounds and costs $500 f. o. b. cars 
at Springfield, Ohio. One man can place this outfit alongside the 
track and perform the welding without interruption to traffic, but 
a helper is usually provided to remove the trolley pole for passing 
cars and to assist generally in the work. The operator himself 
should be a man of intelligence, initiative, and ingenuity, and is 
usually paid from $2.50 to $4 per day. The current strength 
required is from 150 to 180 amperes and the time necessary to 
weld a joint is variously reported at from 30 minutes to over 1 

150211°— 19 8 

H4 Technologic Papers of the Bureau of Standards 

hour. The energy consumption, therefore, based on 175 amperes 
and a time of 45 minutes is 72.2 kw. hrs. per joint on a 550-volt 

The Simplex joint plates, shown in Fig. 11, cost from $1.95 to 
S2.25 and the Apex type of plates, shown in Fig. 12, from $3 to 
$3.50. The welding steel, which is supplied in various grades by 
the company to meet different requirements, is relatively a small 

Where it is necessary to maintain traffic over joints which are 
being welded, the plates are held in place by two bolts temporarily 
installed instead of by the clamps shown in Fig. 19. In order 
to reduce the liability of overheating the rail in any one place 
the following instructions have been issued by the company to all 
purchasers of plates : 

In case of old rails, remove rust, grease, pitch or moisture, using an old file or wire 
brush, and, if necessary, the carbon arc. 

Spot each end of each plate, first on the base line, then on the top line, for a distance 
of about 1 inch, to insure holding the plate in position and to resist the tendency to 
kink or creep during welding. 

Run a heavy fillet of steel around edges of plates, drawing a pocket or cavity in 
both the edge of the plate and the rail. This is important. At the same time deposit 
a globule of the molten electrode, filling in the cavity and building up at least one- 
fourth inch. 

Weld plates as follows (refer to Figs. 11 and 12): (a) Weld on base line from each 
end to center; (6) weld opposite plate the same; (c) return to the first plate and 
weld from the base upward on the sloping cut on each end of the plate; (d) weld both 
ends of opposite plate in similar way; (<?) return to first plate and weld along top of 
plate, beginning at the end and welding to the center; (/) weld top of opposite plate 
in same manner; (g) do not attempt to weld across the undercut at ends of plates, as 
the point is cut off to prevent joining the two lines of welding. 

Some companies are welding standard fish plates or angle bars 
to the base and head of the rail, while still others are using bolted 
or riveted joints and welding only the base, such modifications 
being designed principally to overcome the objection to heating 
the web of the rail. 

That the use of the arc-welded plates is justified from the 
mechanical standpoint alone is shown by the fact that they are 
being used successfully in Cincinnati, where the double-trolley 
system makes rail bonding unnecessary. F. J. Venning, super- 
intendent of construction for the Cincinnati Traction Co., states 

Rail Joints and Bonds 115 

that standard plates are used for this purpose and that they are 
welded along the head of the rail and to a plate or shoe placed 
under the base of the rail. The joint costs about $4 and is said 
by Mr. Venning to be cheaper and better than the continuous 
joint. Incidentally, Mr Venning states that the company is 
making most of its frogs and special work, using old rails and 
the arc welder for the purpose, and is saving as much as 50 per 
cent on some jobs. 

The Dayton, Springfield & Xenia Southern Railway Co. is using 
the electric arc to weld their rail joints to Abbott base plates. 
They also spot weld the nuts on the joints, which prevents them 
from working loose. 

The Springfield Railway Co., of Springfield, Ohio, is introducing 
arc-welded joints on all straight work. Steel ties are used, and 
these are spot welded at all rail ends, thus affording good cross 
bonding. George C. Towle, general manager of the company, 
predicts that all city tracks will be welded within a few years, 
possibly including special work as well. He states that the man- 
ganese steel now being used will last from 10 to 15 years, and can 
therefore be welded the same as straight work, but that its high 
resistance will probably necessitate supplementary bonding. 

The Ohio Electric Railway Co. installed a number of arc- 
welded joints in Springfield, Ohio, more than a year ago, and 
during the first winter a number of rails broke outside of the 
welds. A number of these breaks occurred on rails where the 
pavement had been completed on only one side and are therefore 
attributed to excessive contraction during the cold weather. It 
is also said that the rails have a particularly high carbon content, 
which might account for the high percentage of failures. 

The Columbus Railway, Power & lyight Co., of Columbus, Ohio, 
is using a special joint, which is illustrated in Fig. 20. Special 
plates, which hug the web of the rail with practically no clear- 
ance, are carefully bolted to the rails after reaming the holes for a 
driving fit. These plates are welded to the base of the rail, and 
a short section of a Carnegie steel tie is inverted under the joint 
and also welded to the base of the rail. E. O. Ackerman, engineer 
of way for the company, claims that this joint is not only eco- 


Technologic Papers of the Bureau of Standards 

nomically installed, but that it is giving excellent results, with no 

failures. He does not like the 
idea of heating either the head or 
web of the rail and believes that 
a head support is essential to best 
results. These features he has 
incorporated in the special joint 
here described. 

The Butte Electric Railway 
Co. is using the electric arc to re- 
pair all broken thermit welds and 
also to some extent on regular 
construction. They are not 
afraid to heat the head of the 
rail as is shown in Fig. 21. In 
explanation of this figure the 
company says : 

This drawing shows our method of weld- 
ing this particular joint. With other types 
of rails and joints the manner is varied to 
suit the type. For instance, in welding a 
joint on a 5 2 -pound T rail, angle and bolt 
joint, we use a welding plate which will be 

just wide enough to reach from the flange of the angle bar to the top of the rail, and 

we weld the plate to the top of the rail and to the 

angle bar. We use the carbon for melting or cut- weld ,Wkt& af Joint. 

ting down the rail ends at the joint, afterwards 

filling it up with steel. Of course, on this type 

of joint the angle bar can not be welded to the 

bottom of the rail. 

This drawing does not show the 
longitudinal extent of the weld in the 
head of the rail at the joint, which is 
about 2 inches. 

One of the principal objections 
which has been found to the arc- 
welded joint is that the process is 
slow, requiring about one hour per 
joint. However, as the welding can 
be done without interruption to traffic and requires the time 
of only two men this objection is not serious. The energy 

Fig. 20. — Combination welded and bolted 
joint used in Columbus , Ohio 


Fig. 21. — Combination welded and 
bolted joint used in Butte, Mont. 

Rail Joints and Bonds 117 

consumed is considerable, but even when figured in at a liberal 
rate the total cost of a joint is lower than for most other types 
of welds. 

(e) The Clark Joint. — This joint, as used in Cleveland, and a 
modification of it as used in Baltimore, have been briefly described 
on page — . They resemble the special joint used in Columbus, 
Ohio, and shown in Fig. 20 in that they are a combination of a 
high-grade bolted joint and a welded joint, the thermit shoe 
shown in Fig. 22 taking the place of the weld on the Columbus 

This joint has been standard with the Cleveland Railway Co. 
since 1906, where 50,000 of them are in operation, and where only 
three failures have been recorded. The joint is said to cost from 
$4.60 to $5. 

The modified Clark joint is standard with the United Railways 
& Electric Co., of Baltimore, where they report no failures up to 
the present time. The joint is said to cost from $7 to $9. 

The Clark joint, or one similar to it, has also been installed in 
Buffalo. H. I,. Mack, superintendent of tracks and lines of the 
International Railway Co., states that in installing these joints he 
bonded one rail leaving the other without bonds. Subsequent 
tests showed the unbonded joints to have an efficiency of from 60 
to 70 per cent while the bonded joints showed about 90 per cent. 

While the electrical efficiency of this type of joint is undoubtedly 
less than that of other fully welded types, it is of a permanent 
nature and no other form of bonding should be necessary. 

(/) The Nichols Composite Joint. — This joint, which is 
described on page 22, has had rather a limited application, but 
has nevertheless met with excellent results in Philadelphia and 
St. Louis. Following is an extract from a letter from George B. 
Taylor, engineer of way for the Philadelphia Rapid Transit Co. : 

The Nichols composite rail joint was first used in this city in 1901, and since that 
time practically all of our 9-inch track has been equipped with such joints, and some 
of our T rail has also been so equipped; the total mileage at the present time being 
about 330 miles. All of the track constructed or reconstructed during 19 14 had these 
joints applied, and we expect to apply the same during the coming season. We have 
removed practically none because of defects, but a small amount of track, constructed 
about 1902, and equipped with the Nichols joint, was removed a few years ago for the 
simple reason that the rail was worn out. 

1 1 8 Technologic Papers of the Bureau of Standards 

Over 8000 of these joints are in service in St. Louis where they 
report that no failures have occurred in the three years since their 

The joint is expensive and must be installed by experts who 
realize the necessity of careful work. Tests made in Philadelphia 
by the Bureau of Standards some years ago show that the con- 
ductivity of the joint is practically the same as that of the unbroken 


During the course of its investigation of the subject of rail 
bonds and rail joints the Bureau asked a number of the manufac- 
turers to submit samples of their products for experimental as 
well as for exhibition purposes. They very willingly responded 
and a number of bonded and welded rail joints were collected in 
this manner. These were all tested for conductance and a few 
specimens were tested by the department of- metallurgy of the 
Bureau for the effect of heat on the steel. 

The Bureau of Standards realizes that laboratory tests on 
individual specimens afford no definite indication of the value of 
the average bond or joint under service conditions and therefore 
wishes to caution against placing too much reliance upon the 
results here given. However, the specimens, as a rule, are normal 
and should give a fair idea of what should be expected of new 
bonds and joints of similar types. All resistances were deter- 
mined by comparing them with a Leeds & Northrup 0.000 1 ohm 
shunt, the comparison being made with a high-resistance Weston 
milli voltmeter. A current of about 200 amperes was used in all 

A description of the several bonded and welded joints tested 
and the results obtained are given in Tables 7 and 8. It is of 
interest to note that all of the bonded joints showed a much 
greater conductance with the plates bolted in place than when 
removed. Mention has already been made of the fact that tests 
on old unbonded joints show that a large per cent of them have 
a resistance greater than 1000 feet of rail and that this change 
in the conducting power of joint plates might easily have been 
mistaken in many instances for a deterioration of the bond 

Rail Joints and Bonds 


Resistance Tests on Bonded Rail Joints 

(All resistances given in microhms, double contact resistance taken as the difference between resistance of 
joint across extremities of abutting rails and resistance of bond between terminals.] 

American Steel & Wire Co.'s 































bonds, Electric 
Railway Im- 
provement Co. 



welded bonds 



Brass Co. 



Weight of rail (pounds) 

Length of joint plates (inches) . . . 
Number of bolts in joint plates. . . 

Capacity of bond 

Circuit length of bond (inches) . . 
Distance on rail, center to center 

of terminals (inches) 

Resistance of 3 feet of joint, 

plates in place 

Resistance of 3 feet of joint, 

plates removed 

Resistance across extremities of 

rail ends, plates removed 

Resistance of bond, center to 

center of terminals 

Double-contact resistance 

Single-contact resistance 

Resistance of 3 feet of joint in 

feet of adjacent rail, plates on 
Resistance of 3 feet of joint in 

feet of adjacent rail, plates re- 














































a No plates. b Not taken. 

Resistance Tests on Welded Rail Joints 

Electrically welded joints, 
Lorain Steel Co. 







& Frog 
Co., arc- 

Weight of rail (pounds) 

Type of rail 

Resistance of 3 feet of rail (in microhms) . 
Resistance of 3 feet of joint (in microhms) 
Efficiency of 3 feet of joint (per cent) 


















120 Technologic Papers of the Bureau of Standards 

At the request of one of the manufacturers the Bureau has 
attempted to determine what effect the heat, accompanying the 
welding of bonds or steel plates, has on the grain structure of 
the rail and to what depth a change in grain structure takes place. 
Several specimens were accordingly turned over to the metallur- 
gists of the Bureau, who tested the hardness of the steel at numer- 
ous points by the Brinell test and also made microscopic examina- 
tions of the grain structure in and out of the " weld zone. " 

The Brinell test, which consists in measuring the penetration 
of a cylindrical steel point under a given pressure, showed no appre- 
ciable difference in the hardness of the steel in the "weld zone" 
and at other locations. However, as tests could not be made 
closer than about three-sixteenths inch from the edge of the 
steel by this method, it gave no indication of the hardness immedi- 
ately adjacent to the weld. 

Photomicrographs, together with the report of Dr. Merica, who 
made the examinations, are shown in the accompanying figures. 

Fig. 23, with a magnification of about two diameters, shows 
a portion of the cross section of a rail after being etched, to which 
steel plates had been welded by the electric arc. Figs. 24, 25, 
and 26 show the grain structure close to the weld, at the edge 
of the weld zone, and in the center of the web, respectively. They 
are magnified to about 100 diameters. 

Fig. 27 shows a portion of the head of a rail to which a bond 
had been applied by the oxy-acetylene flame and Figs. 28, 29, 
and 30 show the grain structure close to the weld, at the edge 
of the "weld zone," and outside of the weld zone, respectively. 
Fig. 31 shows a portion of the web of a rail to which a bond had 
been electrically welded, and Figs. 32, 33, and 34 show the grain 
structure close to the weld, at the edge of the "weld zone," and 
in the center of the web. Following is the report of Dr. Merica: 


Transverse sections of parts of the rail adjacent to the weld were polished and 
etched with alcoholic HC1. 

That the structure had been changed by welding could be immediately seen by 
the fact that immediately adjacent to the weld the steel etched much more heavily. 
This is shown in photographs Nos. 322, 321, and 323 of specimens 855, 858, and 857, res- 
pectively (corresponding to Figs. 23, 27, and 31). 

Upon microscopic examination of the structure within and near this "weld zone" it 
was seen that this zone represented that metal which had suffered grain growth or 

Bureau of Standards Tech 

nologic Paper No. 62 

BfJt,.'S/*'- ; ^ ,'•:> 


• ' &I ■ 1 


_ - &>,# HHf 

^B ■ ■ ''*r^ 

f i 




Fig. 25 

Fig. 23. — Heat penetration in arc welding 


Fig. 24 

Fig. 26 

Figs. 24, 25, and 26. — Photomicrographs showing grain-structure of steel after 

arc welding 

Bureau of Standards Technologic Paper No. 62 

Fig. 27. — Heat penetration in oxy-acetylene 

Fig. 29 

Fig. 2\ 

Fig. 30 

Figs. 28, 29, and 30. — Photomicrographs showing grain-structure of steel after 
oxy-acetylene welding 

Bureau of Standards Technologic Paper No. 62 

Fig. 31. — Heat penetration in electric 

Fig. 32 

Fig. 34 

Figs. 32, 2,3, and 34. — Photomicrographs showing grain-structure of steel after 

electric welding 

Rail Joints and Bonds 121 

recrystallization. At the extreme edge of the copper (or steel) the metal had been 
heated above 700 , the transformation point, and had recrystallized, cooling rapidly 
to form a fine-grained structure. Near the edge of the zone the metal had been 
heated just under the transformation point and had merely undergone grain growth, 
as evidenced by the coarser structure at this point. Without this zone no change in 
structure had taken place. The zones vary in depth up to 0.8 cm. 

This change in structure can not be considered serious from the standpoint of the 
wear of the rail. 


Owing to the great number and variety of details involved in the 
foregoing discussions a complete summary of all data and conclu- 
sions will not be attempted. Attention, however, will again be 
called to some of the more important features of the subject under 

Among the most important tendencies as revealed in the in- 
vestigation is the attitude which the companies are now taking 
toward the whole subject of bonding. That it is an engineering 
problem deserving of as much skill and attention as any other 
problem in connection with the operation of an electric railway 
is apparently being realized by the large majority of the railway 
engineers. The testimony given herewith shows a marked tend- 
ency to get away from all types of soldered bonds which, even in 
recent years, have been installed in great numbers. A few com- 
panies who employ thoroughly experienced and careful workmen 
still continue to use them but the number is relatively small. 

Practically all types of standard modern bonds, when selected 
to meet local conditions and installed according to the best 
modern practices, will give satisfactory results with an almost 
negligible percentage of failures on joints which are properly 
maintained. The problem of rail bond maintenance is largely 
that of joint maintenance. No bond can be expected to last 
continuously on a loose and poorly supported rail joint. No one 
type of bond can be said to be better than all other types. Each 
has its advantages and disadvantages and the selection of a bond 
for any particular service should be governed by the type of 
construction on which it is to be used, the grade of labor available 
for installation, and upon numerous other local conditions. 

While welded joints are being used more than ever before there 
is also a growing tendency to adopt improved mechanical. joints 

122 Technologic Papers of the Bureau of Standards 

and various forms of special joints, several of which are a combi- 
nation of welded and bolted or welded and riveted j oints . These spe- 
cial j oints seem to be meeting the demands of service with less failures 
and better results generally than any of the standard types. 

It has been demonstrated that the saving of power alone will 
not justify the best modern practice in bonding. Such practice, 
however, is justified and strongly recommended from the stand- 
point of good voltage conditions in the return circuit, which not 
only make for good electrolysis conditions but also for more satis- 
factory operation. 

Attention is again called to the fact that the problem of track 
bonding is still in a state of evolution. New inventions and 
improvements in methods and practices have been so frequent 
during recent years that many types of bonds and joints can 
still be said to be in the experimental stage. Carefully kept 
records and a free interchange of experiences on the part of the 
operating companies will do much toward the establishment of 
definite and standard practice in this particular field. 

In conclusion we wish to thank the many operating and man- 
ufacturing companies which have contributed the data and 
information contained in this paper. They include not only 
those whose names are formally recorded herein but many others, 
which, through personal interviews and correspondence, have 
rendered no less valuable assistance. 

To Dr. E. B. Rosa and Mr. Burton McCollum, chief physicist 
and electrical engineer, respectively, of the Bureau of Standards, 
is due recognition for conceiving and outlining the scope of this 
paper, as well as for offering many invaluable suggestions regarding 
the collection and arrangement of the material which it contains. 

Special thanks are due to C. S. Kimball, of the Washington 
Railway & Electric Co., who is also chairman of the committee 
on way matters of the American Electric Railway Association, 
and to Prof. Albert S. Richey, of the Worcester Polytechnic Insti- 
tute, for reading this paper in manuscript and offering valuable 
suggestions for its final revision. 

Washington, October 14, 191 5. 


Low, George P. Rail Bonding and its Bearing on Electrolytic Corrosion, A. I. E. 

E., vol. ii, p. 857, 1894. 
Ricker, C W. On Track Bonding, A. I. E. E., Feb. 24, 1905. Vol. 24. 
McMath, T. B. Rail Bonds, Electric Railway Review, Mar. 30, 1907. 
Parshau,, H. F. Earth Returns for Electric Railways, Journal of Institution of 

Electrical Engineers, vol. 27, N. 135, p. 440. 
Harrington, W. E. Rail Bonds, Franklin Institute, 1904. Vol. 157. 
Dm, Mar, W. A. A Much Needed Investigation, Electrical World, Apr. 1, 1907, 

p. 814. 
Hendle, W. A. Rail Bonds, Electric Railway Journal, Apr. 2^, 1910. 
Venning, F. J. Electric Welding in Track Repairs, Electric Railway Journal, Oct. 

10, 1914. 
George, Howard H. The Return Circuit, a series of articles in Electric Railway 

Journal, 1914-15. 
Del Mar, W. A. Electric Power Conductors, D. Van Nostrand, 1914. 
Richey, A. S. Electric Railway Handbook, 1915. 



[Comprising material additional to that in the 1915 edition] 
NOTE. — Special attention should be given to the appendix in connection with the following subjects: 

Brazed or electric-weld bonds 15, 16, 83-86 

Cast weld 19 

Thermit-welded joint 20 and 108 

Electrically welded joint 21 and 111 

Arc-welded joint 112 

Since the issue of this paper in 191 6, there have been no impor- 
tant changes in the art of bonding and welding rail joints, although 
a number of improvements have been made in existing processes. 
One marked tendency has been toward the more general use of 
all types of welded bonds with the almost complete abandonment 
of soldered bonds and those mechanically applied to the head of 
the rail. Pin-terminal and compressed-terminal bonds are still 
extensively used for application to the web of the rail, but even 
here the welded type is finding favor with many companies. One 
reason for the increasing use of gas and electric weld bonds is to 
be found in the development of lighter, cheaper, and less unwieldy 
tools for their application. Some of the newer methods and appa- 
ratus which have been developed for this class of work are so far 
superior to those formerly employed that it seems appropriate to 
include a description of them in the second edition. 

With the abnormal price of copper during the war period, the 
cost of long bonds became almost prohibitive and this also acted 
as a stimulus to the use of short head bonds. Although little 
new track work of a permanent character was done during the 
war, there has been some advancement in the welding of rail 
joints and the most important improvements in this field will also 
be described here. 

(a) Electric- Weld Bonds. — A marked improvement over the 
bonding car shown in Fig. 6 and described on page 86 has been de- 
veloped by the Electric Railway Improvement Co., and is known 




Bureau of Standards Technologic Paper No. 62 

Fig. 36. — Web welder 

Rail Joints and Bonds 125 

as the Erico portable welder. It produces the same weld as was 
formerly obtained with the larger and more expensive car and in 
addition may be used as an arc-welding set for work either in the 
shop or along the track. The new apparatus shown in Fig. 35 
consists of the welder proper and a rheostat, the combined weight 
of which is said to be less than 250 pounds. 

The welder part of the equipment consists of an inclosed furnace 
box on the front side of which is inserted a graphite plate which is 
brought in contact with the rail bond terminal. This furnace box 
is surrounded with a magnetic winding, which is in series with the 
heating current. Through the center of this winding passes a car- 
bon electrode which enters the back of the furnace box through a 
narrow slot and makes contact with the graphite front plate. The 
process of welding the bond terminal consists in establishing an arc 
in the furnace box between the carbon electrode and the graphite 
front plate which presses against the bond terminal. In producing 
the weld the arc does not come in contact with either the bond or 
rail, thus avoiding the danger of injury to the bond, rail, and eyes 
that such methods are liable to introduce. 

It is claimed that the inclosed magnetic arc of the welder has a 
potential drop of about 125 volts and requires from 100 to 125 
amperes; taking about one minute to weld a No. 4-0 terminal to 
the rail. The winding has the effect of focusing the arc directly on 
the front plate where the heat intensity is required. 

The mechanical details of the welder are such that the appa- 
ratus is at no time attached to the rail; it can therefore be re- 
moved from interference with traffic by simply picking up and 
setting to one side. It is supported in operating position at two 
points on the rail to be bonded and by a hook reaching over to the 
other rail. Adjustment of the contact of the graphite front plate 
against the bond terminal is effected by varying the length of the 
hook and by raising or lowering the body of the welder by means 
of a vertical screw operated by a hand wheel. 

The welder, as used for installing bonds underneath the fish- 
plates and for attaching cables around the plates, is shown in 
Fig. 36 and weighs approximately 40 pounds. 

Current is supplied to the welder directly from the trolley wire 
through a special portable rheostat, which is designed to secure a 

126 Technologic Papers of the Bureau of Standards 

wide range in current capacity, with the maximum radiating sur- 
face and the minimum weight of materials used. The rheostat is 
composed of a number of independent resistance units in parallel 
producing a range of from 60 to 200 amperes in 15 ampere steps. 
While this range in current is not required in the operation of the 
welder, it is quite essential in using the rheostat for general arc 
welding work. The weight of the rheostat is 140 pounds when 
equipped with flanged wheels for rolling along the track. It is not 
necessary to have the rheostat on the track while the welding 
process is performed — it may be set at one side of the track if 

The rheostat of the portable welder may also be used for install- 
ing arc-weld bonds to the rails by a method of welding which is 
and has been universally used for years in all kinds of arc -welding 

The terminals of the rail bonds used in this process are provided 
with an iron casing, which is electrically welded to the copper in 
the process of manufacture. 

The arc-weld bond is placed against the rail for welding and is 
so adjusted that a V-shaped space is produced between the rail 
and iron casing of the bond. This space is then filled with weld- 
ing rod by means of the electric arc. The attaching of the iron- 
clad bonds is therefore an operation similar to that with which all 
arc welders are familiar, and which any arc welder should do with 
ease, and one which an inexperienced employee may readily be 

The results obtained by welding iron to iron are said to be much 
more satisfactory than where copper is welded to iron owing to 
the tendency of melted copper to oxidize rapidly, thereby pro- 
ducing a spongy structure having little mechanical strength. F01 
this reason, as well as on account of the greater economy, steel 
terminals or ironclad terminals are commonly used on gas and 
electric weld bonds. 

Another type of electric-weld bond which has come into exten- 
sive use since the publication of the first issue of this paper is the 
L-V type of the Lincoln Bonding Co. This bond, illustrated in 
Fig. 37, is of the laminated, short, exposed type, being 6 inches in 
length and only 2]/ 2 inches between terminal centers. The weld- 

Bureau of Standards Technologic PaDer No. 62 

Fig. 37. — 4-0 Lincoln bond welded to rail 

Bureau of Standards Technological Pacer No. 62 

Fig. 38. — Dynamotor for track bonding 

Rail Joints and Bonds 127 

ing process is essentially different from other similar operations. 
The carbon-arc type of welding is used and the bond is held in 
place prior to and during the welding in a mold of carbon with a 
copper plate forming the base. An arc is struck and the head of 
the bond is melted down and allowed to run over the copper plate, 
after which No. 4 scrap copper is fed into the arc and a solid head 
is built up. An important feature in this welding process is the 
manner in which the arc current is generated and controlled. In- 
stead of being supplied from the trolley wire through a resistance, 
it is produced by a specially designed motor-generator set in one 
unit, called a dynamotor. This machine, together with the neces- 
sary controlling and stabilizing apparatus mounted on it is illustra- 
ted in Fig. 38. The connections are so arranged in this dynamotor 
that the current flows from the bond or the rail to the carbon pencil 
or in the opposite direction to the flow when current is taken directly 
from the trolley. A great advantage is claimed by the manufac- 
turers in this feature on account of the well-known fact that more 
heat is developed at the positive than at the negative terminal of 
an electric arc. 

like all welding operations the character of the work depends 
largely upon the skill and care of the operator. When properly 
applied the contact resistance is very low and this combined 
with the short length of the bonds gives a joint of very low resist- 
ance. Moreover, the small amount of metal in the bond reduces 
the liability of theft. On the other hand, such a short bond is 
not well adapted to loose joints as the continual vibration incident 
to traffic will eventually crystallize the laminations, thus leading 
to the complete failure of the bond. 

The dynamotor is so designed that it is adaptable to metallic 
electrode type of welding and can therefore be used in the shop 
for general repair work. A stabilizing inductance is furnished for 
this class of work which is mounted on the side of the machine. 

Another type of bond which has been developed by this manu- 
facturer is a cable bond which is applied in a horizontal position 
under the base of the rail. To apply this bond it is necessary to 
burn two small notches out of the base of the rail which are filled 
up with copper during the course of the application. This cable 
type of bond is advocated for paved-in track where the usual head 

128 Technologic Papers of the Bureau of Standards 

bond is subjected to forces which tend to break it from the rail. 
This type of bond has also found favor in mines where a head bond 
is likely to be sheared off by dump-car derailments which are 
said to be frequent occurrences. 

(b) The Cast- Weld Rail Joint. — While the cast- weld joint 
has been superseded in most cities by various forms of improved 
electric or thermit welded joints, it is interesting to note that in 
a few cities the results obtained were such as to justify their con- 
tinuous use down to this date, among these cities being Minne- 
apolis, Detroit, and Milwaukee. This joint is also being largely 
used in Brooklyn. In Milwaukee there are many cast-weld joints 
which are 20 years old, and which are in such good condition that 
their location in the track can hardly be distinguished on inspec- 
tion. A new type of cast- weld joint has recently been developed 
in Milwaukee which promises to give much better results than the 
old rectangular form of weld which was formerly used and in 
which many failures developed. In the new joint by keeping the 
weld well below the head of the rail and taking precautions to keep 
the head of the rail cool by means of a water-filled strong back 
while pouring the weld, the running portion of the rail retains its 
original properties and does not have so great a tendency to cup 
under continuous traffic. Fig. 39 shows two of these joints after 
having been tested to destruction. They were made from 7-inch, 
9 5 -pound T rail and the pressure was applied midway between 
supports, 5 feet apart. The joints failed at 50 tons and 54 tons, 
respectively. It will be noted that in both cases the failure occurred 
in the rail and not in the weld. Similar joints failed under a tensile 
stress of 240 tons with steel bond plugs and 210 tons without 
steel bond plugs. These tests compare favorably with similar 
tests on other types of welded joints. The cast weld in this 
altered form is considered as the standard in Milwaukee, although 
a considerable percentage of repair joints are now being made 
by other methods. 

(c) Thermit- Welded Rail Joints. — The present improved 
form of this joint referred to on pages 20 and 109, but not described, 
is known as the thermit fully welded rail joint. 

In making this weld the rails are spaced three-fourths of an 
inch apart, in the case of new rail, and a gap of that width is 

Bureau of Standards Technologic Paper No. 62 

Fig. 39. — Improved cast weld rail joints tested to destruction 

Bureau of Standards Technologic Paper No. 62 


Fig. 40. — Thermit insert fully welded rail joint 

Rail Joints and Bonds 129 

provided by cutting or sawing in the case of old rail. Into this 
space and between the heads of the rails only is placed an insert 
cut from a rolled section of similar analysis to the rail itself. 
These inserts are provided in a variety of thicknesses to facilitate 
fitting snugly and the top surface is allowed to project slightly 
above the surface of the rail head. The accompanying illustra- 
tion, Fig. 40, shows the position of the insert and how the thermit 
steel is flowed around the joint to make the weld. The web and 
base are thoroughly melted and amalgamated with the thermit 
steel but the entire head of the rail is not. Only the outside of 
the head and the lip of the rail are melted by the thermit steel. 
A weld of the entire rail section is obtained, however, when the 
thermit steel begins to cool and contract. The proximity of this 
steel to the insert has melted up the lower part of the insert and 
heated the upper part as well as the abutting heads of the rails 
to such a high temperature that when contraction sets in, thus 
drawing the rails together with tremendous force, the effect is 
to butt weld the insert permanently into place between the rail 

The advantage claimed for this method is to be found in the 
fact that the metal in the head of the rail coming into contact 
with the car wheels is not melted and its physical properties not 
changed. As a result, it is said, welds made in this way have 
shown no tendency to cup after six years of hard service under 
heavy and frequent traffic. Furthermore, the breakage of welds 
made in this way has proved to be so small in the six years since 
the welds have been installed that they can hardly be figured in 
percentages at all. The manufacturers state that in San Antonio 
where nearly 10 000 welds have been made since 191 3, the number 
of broken joints has been negligible, and at last reports the number 
of breaks had averaged less than 1 in 2000 welds. It is claimed 
that similar reports have been received from other cities where 
the joint has been installed for several years. Where breaks 
have occurred they have usually taken place either shortly after 
welding, indicating something wrong with the procedure, or else 
they have occurred during the first winter. 

An interesting development in the use of thermit for rail welding 
has been its application to the welding of special work; that is, 
150211°— 19 9 

130 Technologic Papers of the Bureau of Standards 

making up frogs and crossings by welding together short pieces of 
rail. The first crossing to be made up in this manner was welded 
by the Milwaukee Electric Railway & Light Co., Milwaukee, Wis., 
in 1 914, and it is said to have been in good condition in August, 
1 919. Other companies using the process in the same way are 
the Indianapolis Traction & Terminal Co., who have been doing 
it for several years and the Omaha & Council Bluffs Street Railway 
Co., which has recently taken it up. Fig. 41 shows a piece of 
thermit special work installed in Indianapolis. 

It is said that the thermit-insert weld seems to be the only 
process which will successfully weld alloy steel rails such as the 
Mayari rails made by the Bethlehem Steel Co. Other methods 
have not proved successful in welding these rails but 500 welds 
were made in Chicago by the thermit-insert process and only 3 
joints have broken since. The welds were made in 191 5 while 
the 3 breaks occurred in the winter of 191 6. These were rewelded 
and no further breaks have occurred. 

(d) The Electrically Weeded Joint.— The electrically welded 
joint with spelter head support, shown in Fig. 10, and made by 
the Lorain Steel Co. has been superseded by a chock head sup- 
port bar joint shown in Fig. 42. This joint was introduced in 
1 91 6. A drop-forged steel chock is welded over the bars at the 
center to provide a head support for the abutting rails. A weld 
is made between the back of the rail head and the chock as well 
as above the bars. Fig. 43 shows an electrically welded chock 
joint in which the use of the bars has been dispensed with. These 
joints were first used in Chicago in 191 8. 

(e) Arc- Welded Joints, — A comparatively new type of arc- 
welded joint is the so-called Gailor joint exploited by the Lincoln 
Bonding Co. This joint was first put on the market by the 
Atlantic Welding Corporation and was known as the Atlantic 
joint. The joint is formed by welding plates, which stand away 
from the web and fish to head and base, along their top and 
bottom edges for the full length of the plates. The principal 
feature of this joint is that the plates support the head on both 
sides for the full length of the plates. If properly welded, the 
conductivity of this joint is in excess of that of the rail section. 
The joint can be made by either using mechanical fishplates on 



Bureau of Standards Technologic Paper No. 62 

Fig. 44. — Gailor arc-welded joint 

. -«HHHHI 

Fig. 45. — Dynamotor in use for joint welding in the street 

Rail Joints and Bonds 131 

hand, beveled in accordance with the manufacturer's instructions, 
or by the railways making plates from bar steel. Special plates 
may also be obtained from the Rail Joint Co. Fig. 44 shows 
a welded joint made with flat bar type plates. 

A special process of welding is usually employed for welding 
these joints which makes it possible to conveniently weld the 
plates to the under side of the rail head. The process is a carbon- 
arc method in which the welding is performed in the presence of a 
composition bar high in copper. It is claimed that this bar has 
a stabilizing effect on the arc and permits of obtaining a good 
weld with deep penetration at the rate of 6 linear inches per 
minute. The parts necessary for this welding have been so 
designed and developed that a minimum amount of equipment 
is necessary. 

The Lincoln Bonding Co. has developed a larger size of their 
dynamotor weighing about 1300 pounds for the supply of current 
for this process. This machine is shown in Fig. 45. The accesso- 
ries and especially the clamps for holding the copper-composition 
bars have been especially designed to permit welding under traffic 
and so as not to interfere with other vehicular traffic on the 
street. It is claimed that with this equipment a crew of 3 men 
can weld 25 joints per day in new track construction without 
traffic or 14 joints per day under 3-minute traffic in 7^ hours. 
The dynamotor of this equipment has also been developed to 
perform metallic electrode welding for building up cupped rails, 
for welding rail joints in worn track, and for repairing special 
work. Like the smaller bonding machine the welding generator 
is equipped with a stabilizer which is mounted on the side of the 

Washington, October 10, 1919.